Driving control method of solar tracking system

Abstract

The invention discloses a driving control method of a solar tracking system, which belongs to the technical field of solar tracking systems and comprises the following steps: acquiring historical working condition data of the solar tracking system and illumination tracking data corresponding to the historical working condition data; illumination tracking of the solar tracking system is simulated and trained under different working conditions, and a trained sunlight tracking reinforcement learning model is obtained; updating control parameters of a PID controller used for controlling and driving the azimuth angle steering engine and the elevation angle steering engine to obtain target PID control parameters; acquiring real-time working condition data of the solar tracking system; and according to the real-time working condition data of the solar tracking system, driving and controlling the solar tracking system to perform illumination tracking based on the trained sunlight tracking reinforcement learning model and the target PID control parameters. The problem that an existing solar tracking system is insufficient in illumination tracking accuracy is solved.

Description

Technical Field

[0001] The present invention belongs to the technical field of solar tracking systems, and in particular relates to a driving control method for a solar tracking system. Background Art

[0002] Solar energy is a renewable, clean and abundant energy source. In the field of photovoltaic power generation technology, insufficient power collection caused by low power generation efficiency is a major problem at this stage. This is related to the fact that the earth revolves around the sun in a periodic motion and rotates around its axis, causing the relative position between the sun and the earth to constantly change. The solar tracking system is a device that can automatically track the direction of the sun's movement. Its goal is to ensure that solar light always shines on the solar panel at the best angle, thereby effectively improving the power generation efficiency of the solar panel.

[0003] The principle of photoelectric tracking in solar tracking system is: using photosensitive device to sense the sun's light, as the angle of illumination changes, the photosensitive device generates differential analog signal, the differential analog signal can be converted into electrical signal through signal conversion circuit, the electrical signal is transmitted to the controller of solar tracking system, after calculation and processing, the controller issues control instruction to drive the solar tracking system to perform corresponding movement, so that the solar panel receives sunlight at the best angle. Although the photoelectric tracking method realizes a certain degree of closed-loop solar tracking through electrical signal transmission, it is difficult to accurately judge the position of sunlight under complex working conditions, which can easily lead to large positioning error of sunlight, and the deflection of solar panel is difficult to ensure accurate drive, thus affecting the light tracking effect. Summary of the invention

[0004] In view of the above-mentioned deficiencies in the prior art, the present invention provides a driving control method for a solar tracking system, which solves the problem of insufficient light tracking accuracy of the prior solar tracking system.

[0005] In order to achieve the above-mentioned object of the invention, the technical solution adopted by the present invention is:

[0006] The present invention provides a driving control method for a solar tracking system, comprising the following steps:

[0007] S1. Obtain historical operating data of the solar tracking system and light tracking data corresponding to the historical operating data;

[0008] S2. According to the reinforcement learning method, the historical working condition data of the solar tracking system and the corresponding light tracking data, the light tracking of the solar tracking system is simulated and trained under different working conditions to obtain a trained sunlight tracking reinforcement learning model;

[0009] S3, according to the historical working condition data of the solar tracking system and the corresponding light tracking data, the control parameters of the PID controller used to control the azimuth servo and the altitude servo are updated to obtain the target PID control parameters;

[0010] S4, obtaining real-time working data of the solar tracking system;

[0011] S5. According to the real-time working condition data of the solar tracking system, based on the trained sunshine tracking reinforcement learning model and target PID control parameters, the solar tracking system is driven and controlled to perform light tracking.

[0012] The beneficial effects of the present invention are as follows: a driving control method for a solar tracking system provided by the present invention simulates and learns the light tracking data of a solar tracking system under different working conditions according to a reinforcement learning method and a reward mechanism of reinforcement learning, so that a trained sunshine tracking reinforcement model can be obtained when the reward threshold is greater than that for different working conditions. The trained sunshine tracking reinforcement model can obtain predicted sunshine tracking data that meets the accuracy requirement, and provide a data basis for accurately driving and controlling the azimuth servo and the altitude servo based on the predicted sunshine tracking data; the present invention provides a method for updating the control parameters of a PID controller for controlling the azimuth servo and the altitude servo according to different historical working condition data of the solar tracking system and the corresponding sunshine tracking data, and obtains the target PID control parameters. Based on the target PID control parameters, the azimuth servo and the altitude servo can be accurately driven and controlled to deflect by using a PID controller based on the sunshine tracking data, so as to continuously and accurately track the solar light and keep the sunlight incident vertically for efficient power generation; the present invention can timely and accurately drive and control the solar tracking system for light tracking through the trained sunshine tracking reinforcement learning model and the target PID control parameters.

[0013] Furthermore, the historical operating data of the solar tracking system in S1 includes light data, temperature data, air humidity data, wind speed data, weather data and power grid data;

[0014] The illumination data includes the solar altitude angle and the solar azimuth angle; the weather data includes sunny, cloudy, rainy and snowy weather; the power grid data includes frequency and voltage;

[0015] The light tracking data includes solar panel retraction and expansion data and dual-axis deflection data. The solar panel retraction and expansion data is state data of the solar panel in a retracted state or an expanded state. The dual-axis deflection data includes horizontal angle deflection data and altitude angle deflection data.

[0016] The beneficial effect of adopting the above further scheme is: the present invention fully considers aspects such as light, temperature, air humidity, wind speed, weather and power grid for the operating data of the solar tracking system, which are closely related to the power generation efficiency and power generation status of the solar tracking system. Based on comprehensive operating condition considerations, it can effectively improve the accuracy of light tracking data prediction for the solar tracking system, and can also improve the accuracy of the target PID control parameters obtained through training.

[0017] Furthermore, the calculation expressions of the solar altitude angle and azimuth angle are as follows:

[0018] sin Hs=sin°S sinσ+cos°S sinθ s ,

[0019] θ s =(t z -12)×15°, t z =t p +Δ z , t p =t bj +t z ,

[0020]

[0021] Δ z (rad)=0.000766+0.001869cosθ0+0.014616cos2θ0-0.04085sin2 θ ,

[0022]

[0023] Among them, Hs represents the solar altitude angle, °S represents the local latitude, σ represents the declination angle, θ s represents the solar hour angle, As represents the solar azimuth, σ (rad) represents the radian corresponding to the declination angle, N represents the number of days that have passed in the year, and t z represents true solar time, t p represents mean solar time, Δ z Indicates the true solar time difference, t bj Indicates Beijing time, t xz represents the accuracy correction time, °E represents the local longitude, Δ z (rad) represents the radian corresponding to the true solar time difference, θ0 represents the day angle, and the solar azimuth is the azimuth of 0° in the south direction.

[0024] The beneficial effect of adopting the above further scheme is: the present invention provides a method for calculating the solar altitude angle and azimuth angle, which can accurately calculate the solar altitude angle and azimuth angle based on the longitude and latitude position information and the current time information, and can provide a basis for accurately predicting the dual-axis deflection data in the light tracking data.

[0025] Furthermore, S2 comprises the following steps:

[0026] S21. According to the reinforcement learning method, the network architecture and initial network parameters of the sunshine tracking reinforcement learning model in the current round of training are set;

[0027] S22, inputting the historical operating condition data of the solar tracking system into the light tracking reinforcement learning model in sequence for simulation training, and obtaining predicted light tracking data corresponding to the historical operating conditions;

[0028] S23, calculating the reward for the current round of training based on the reward model according to the predicted light tracking data corresponding to different historical working conditions;

[0029] S24. According to the reward of the current round of training, the network parameters of the sunshine tracking reinforcement learning model are adjusted to obtain the initial network parameters for the next round of simulation training;

[0030] S25. Based on S21-S24, several rounds of iterative training are performed until the training rewards corresponding to each historical working condition except rainy and snowy weather are greater than the preset reward threshold, and the training is stopped to obtain a trained sunshine tracking reinforcement learning model.

[0031] The beneficial effect of adopting the above further scheme is: the present invention provides a method for simulating the light tracking of a solar tracking system under different working conditions. Through the reward mechanism, it can ensure that the predicted light tracking data corresponding to each working condition except rainy and snowy weather can meet the light tracking accuracy of the solar tracking system, so as to achieve a photovoltaic power generation effect that meets the needs.

[0032] Furthermore, the calculation expression of the reward model in S23 is as follows:

[0033] Rcur i =(1-ε(t))×Rbase i +ε(t)×Rpbr,

[0034] Among them, Rcur i represents the reward of the i-th historical condition in the current round of training, ε(t) represents the preset dual-axis tracking time function, and Rbase i represents the basic training reward for the i-th historical condition, R pbr Represents the prediction gain reward.

[0035] The beneficial effect of adopting the above-mentioned further scheme is as follows: the present invention provides a calculation method of the reward model, which not only takes into account the different training basic rewards corresponding to light tracking under different working conditions, but also takes into account the predicted gain reward corresponding to the difference between the light tracking data and the predicted light tracking data, and the reward result changes linearly with the time of dual-axis tracking, and can achieve accurate differentiated rewards for the light tracking effects of different working conditions at different times, thereby ensuring that the predicted light tracking data obtained by the sunlight tracking reinforcement learning model under each working condition can ensure the photovoltaic power generation effect that meets the requirements.

[0036] Furthermore, the training basic reward in S23 is related to the illumination data, temperature data, air humidity data, wind speed data, weather data and power grid data during each round of training, and different and fixed training basic rewards correspond to different working conditions;

[0037] The predicted gain reward is correlated with the difference between the light tracking data corresponding to the historical working condition data and the predicted light tracking data in each round of training, which is reflected in the following two aspects: on the one hand, under the same historical working condition data, the more power the solar panel generates and the smaller the difference between the light tracking data and the predicted light tracking data, the greater the predicted gain reward; on the other hand, under the same historical working condition data, the more power the solar panel generates and the greater the difference between the light tracking data and the predicted light tracking data, the smaller the predicted gain reward.

[0038] The beneficial effect of adopting the above further scheme is: under different working conditions in the present invention, there are differences in the corresponding training basic reward value for each working condition. Combined with the predicted gain reward related to the power generation of the solar panel, the difference between the light tracking data and the predicted light tracking data can be effectively reduced, so that the finally trained sunlight tracking reinforcement learning model can predict the predicted light tracking data under the optimal photovoltaic power generation condition for different working conditions, that is, the dual-axis deflection data corresponding to when sunlight directly hits the solar panel.

[0039] Furthermore, S3 comprises the following steps:

[0040] S31, setting initial control parameters of a PID controller for controlling the azimuth servo and the altitude servo, wherein the control parameters include a proportional gain, an integral time constant and a differential gain;

[0041] S32, obtaining the deviation between the dual-axis deflection data under different working conditions and the expected dual-axis deflection data according to the historical working condition data of the solar tracking system and the corresponding light tracking data;

[0042] S33, constructing a control law model and a PID transfer model of the PID controller according to the initial control parameters of the PID controller and the deviation between the dual-axis deflection data under different working conditions and the expected dual-axis deflection data;

[0043] S34, based on the control law model and PID transfer model of the PID controller, update the control parameters of the PID controller until the deviation between the dual-axis deflection data under different working conditions and the expected dual-axis deflection data is less than a preset deviation threshold, and obtain the target PID control parameters.

[0044] The beneficial effect of adopting the above-mentioned further scheme is as follows: the present invention provides a method for training and updating the control parameters of the PID controller through different working condition data and corresponding light tracking data, with the goal of reducing the deviation between the dual-axis deflection data and the expected dual-axis deflection data under different working conditions, and continuously updating the PID controller. It can effectively ensure that after accurately obtaining the predicted light tracking data corresponding to the real-time working condition, it can accurately control the driving direction angle servo and the altitude angle servo to perform precise deflection based on the dual-axis deflection data in the predicted light tracking data, so that the solar tracking system can continuously track direct sunlight and achieve high-efficiency photovoltaic power generation.

[0045] Furthermore, the calculation expressions of the control law model and the PID transfer model of the PID controller in S33 are as follows:

[0046]

[0047] Where u(t) represents the output value of the PID controller driving the azimuth and elevation servos, K p represents the proportional gain, e(t) represents the deviation between the dual-axis deflection data under different working conditions and the expected dual-axis deflection data, T i represents the integral time constant, dt represents the differential with respect to time, It represents the differential of the deviation between the biaxial deflection data under different working conditions and the expected biaxial deflection data with respect to time, T d Denotes the differential gain, K i Indicates the integral coefficient, K d represents the differential coefficient, G(s) represents the transfer function of the PID controller, and s represents the open-loop frequency.

[0048] The beneficial effect of adopting the above further scheme is as follows: the present invention provides a calculation method for a control law model and a PID transfer model of a PID controller. Based on the control law model and the PID transfer model of the PID controller, the control parameters of the PID controller can be updated so that the PID controller can accurately drive the control direction angle servo and the altitude angle servo to perform precise deflection, thereby ensuring that the solar tracker accurately tracks direct sunlight and maintains high-efficiency photovoltaic power generation.

[0049] Furthermore, the S5 comprises the following steps:

[0050] S51, inputting the real-time working condition data into the trained sunshine tracking reinforcement learning model to obtain the predicted sunshine tracking data corresponding to the real-time working condition;

[0051] S52, if the real-time weather data is rainy or snowy, and the solar panel expansion and contraction data in the predicted light tracking data corresponding to the real-time working condition is that the solar panel is in a contracted state, then the solar panel of the solar tracking system is contracted until the weather data is sunny or cloudy, then the solar panel is expanded, and the process proceeds to S53;

[0052] S53, according to the dual-axis deflection data in the predicted light tracking data corresponding to the real-time working condition, based on the target PID control parameters, use the PID controller to control the driving direction angle servo and the altitude angle servo to deflect, so as to control the solar tracking system to perform light tracking.

[0053] The beneficial effect of adopting the above further scheme is as follows: based on the real-time working condition data, the present invention uses the trained sunshine tracking reinforcement learning model to accurately predict the light tracking data. If it is rainy or snowy weather, the solar panel is driven to shrink to prevent the accumulation of rain and snow or the rust of components caused by rain. If it is not rainy or snowy weather, based on the target PID control parameters, the PID controller is used to control the azimuth servo and altitude servo in the solar tracking system to accurately deflect according to the predicted dual-axis deflection data, thereby realizing direct tracking of sunlight and always performing photovoltaic power generation in a high-efficiency state.

[0054] Other advantages of the present invention will be analyzed in more detail in subsequent embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0056] Figure 1 4 is a flow chart of the steps of a driving control method of a solar tracking system in an embodiment of the present invention. DETAILED DESCRIPTION

[0057] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. The components of the embodiments of the present invention generally described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents the selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making creative work belong to the scope of protection of the present invention.

[0058] like Figure 1 As shown, in one embodiment of the present invention, the present invention provides a driving control method for a solar tracking system, comprising the following steps:

[0059] S1. Obtain historical operating data of the solar tracking system and light tracking data corresponding to the historical operating data;

[0060] The historical operating data of the solar tracking system in S1 includes light data, temperature data, air humidity data, wind speed data, weather data and power grid data;

[0061] The illumination data includes the solar altitude angle and the solar azimuth angle; the weather data includes sunny, cloudy, rainy and snowy weather; the power grid data includes frequency and voltage;

[0062] In this embodiment, the solar altitude angle and solar azimuth angle can be calculated based on the longitude and latitude of the solar tracking system, the date of the day and the time. The temperature, air humidity and wind speed of the environment in which the solar tracking system is located can be monitored by temperature and humidity sensors and wind speed sensors. Weather information is remotely received through wireless communication, and combined with rainfall and snowfall meteorological sensor monitoring, weather data in the environment in which the solar tracking system is located can be obtained. The voltage and frequency of the power grid environment in which the solar tracking system is located can be obtained through the intelligent power sensor.

[0063] The calculation expressions of the solar altitude angle and azimuth angle are as follows:

[0064] sin Hs=sin°S sinσ+cos°S sinθ s ,

[0065] θ s =(t z -12)×15°, t z =t p +Δ z , t p =tbj+ t xz ,

[0066]

[0067] Δ z (rad)=0.00076+0.001869cosθ0+0.014616cos2θ0-0.04085sin2θ0,

[0068]

[0069] Among them, Hs represents the solar altitude angle, °S represents the local latitude, σ represents the declination angle, θ s represents the solar hour angle, As represents the solar azimuth, σ (rad) represents the radian corresponding to the declination angle, N represents the number of days that have passed in the year, and t z represents true solar time, t p represents mean solar time, Δ z Indicates the true solar time difference, t bj Indicates Beijing time, t xz represents the accuracy correction time, °E represents the local longitude, Δ z (rad) represents the radian corresponding to the true solar time difference, θ0 represents the day angle, and the solar azimuth is the azimuth of 0° in the south direction.

[0070] The greater the proportion of direct light in the illumination, the greater the increase in the power generation of the solar tracking system, and the more electricity generated by the photovoltaic modules, which can effectively improve the overall efficiency of the solar tracking system; the increase in temperature will cause the output power of the photovoltaic modules to decrease, and will also cause the performance of the electronic components inside the solar tracking system, such as the inverter, to decline, affecting the life of the whole machine. The power generation of the solar tracking system in a high temperature environment will be affected; the increase in air humidity means that the water vapor content in the air increases, and the weakening of solar radiation increases, which will cause the photovoltaic modules to produce a potential induced attenuation effect, thereby reducing the power generation efficiency of the solar tracking system; the wind speed is inversely proportional to the temperature of the components of the solar tracking system. The greater the wind speed, the better the ventilation performance between the components of the solar tracking system, and the lower the component temperature, which can effectively improve the power generation efficiency of the solar tracking system, but in the case of wind In areas with severe sandstorms, higher wind speeds will cause dust to accumulate on solar panels, resulting in a sharp drop in the power generation efficiency of the solar tracking system. Sunny days are most conducive to the operation of the solar tracking system, and the power generation is higher. When the weather is cloudy, direct sunlight is reduced, but the solar tracking system can still receive scattered solar light, and the power generation efficiency is lower than that on sunny days. When it is rainy and snowy, rain and snow block the sunlight, significantly reducing the power generation of photovoltaic modules. At the same time, it may also cause erosion to solar panels, which require protection. The grid voltage may cause voltage or frequency fluctuations after the solar tracking system is connected. For example, when a large-scale solar tracking system generates electricity, it will cause the grid voltage to increase, requiring voltage regulation and transformation. When there is insufficient light, the intermittent and volatile nature of the solar tracking system may cause the grid frequency to drop, requiring energy storage systems and other energy sources to smooth it out.

[0071] The light tracking data includes solar panel retraction and expansion data and dual-axis deflection data. The solar panel retraction and expansion data is state data of the solar panel in a retracted state or an expanded state. The dual-axis deflection data includes horizontal angle deflection data and altitude angle deflection data.

[0072] In this embodiment, the solar tracking system is a dual-axis tracking system, the dual axes are the azimuth control axis and the altitude control axis, the dual-axis structure is convenient for vertically receiving sunlight, and its solar panel can be retracted and unfolded to cope with rainy and snowy weather; the horizontal angle deflection data corresponds to the azimuth servo drive control, and the altitude deflection corresponds to the altitude servo drive control; in this embodiment, the azimuth servo of the solar tracking system is used to drive its solar panel to rotate towards the sun within a full range of the horizontal direction, the output shaft of the azimuth servo points in the same direction as the azimuth control axis, and the azimuth servo is deflected to achieve continuous tracking of the azimuth by the solar panel; the altitude servo of the solar tracking system is used to drive its solar panel to rotate towards the sun within a range of 180° in the vertical direction, the output shaft of the altitude servo points in the same direction as the altitude control axis, and the altitude servo is deflected to achieve continuous tracking of the altitude by the solar panel; in this embodiment, both the azimuth servo and the altitude servo use servo motors to provide a basis for precise deflection.

[0073] S2. According to the reinforcement learning method, the historical working condition data of the solar tracking system and the corresponding light tracking data, the light tracking of the solar tracking system is simulated and trained under different working conditions to obtain a trained sunlight tracking reinforcement learning model;

[0074] The S2 comprises the following steps:

[0075] S21. According to the reinforcement learning method, the network architecture and initial network parameters of the sunshine tracking reinforcement learning model in the current round of training are set;

[0076] S22, inputting the historical operating condition data of the solar tracking system into the light tracking reinforcement learning model in sequence for simulation training, and obtaining predicted light tracking data corresponding to the historical operating conditions;

[0077] In this embodiment, during simulation training, different historical working conditions are all possible states of the solar tracking system; the solar panel contraction, solar panel expansion and azimuth servo deflection, and solar panel expansion and altitude servo deflection corresponding to the light tracking data are the actions of the solar tracking system; the state transfer occurs after the action is executed under the current working condition.

[0078] S23, calculating the reward for the current round of training based on the reward model according to the predicted light tracking data corresponding to different historical working conditions;

[0079] The calculation expression of the reward model in S23 is as follows:

[0080] R cur =(1-ε(t))×R base +ε(z)×R pbr ,

[0081] Among them, R cur represents the reward of the current round of training, ε(t) represents the preset dual-axis tracking time function, R base represents the training base reward, R pbr Indicates the predicted gain reward. The dual-axis tracking time function preset in this embodiment has the characteristic of decaying with time, such as negative linear time decay.

[0082] The training basic reward in S23 is related to the illumination data, temperature data, air humidity data, wind speed data, weather data and power grid data during each round of training. Different and fixed training basic rewards correspond to different working conditions.

[0083] The predicted gain reward is correlated with the difference between the light tracking data corresponding to the historical working condition data and the predicted light tracking data in each round of training, which is reflected in the following two aspects: on the one hand, under the same historical working condition data, the more power the solar panel generates and the smaller the difference between the light tracking data and the predicted light tracking data, the greater the predicted gain reward; on the other hand, under the same historical working condition data, the more power the solar panel generates and the greater the difference between the light tracking data and the predicted light tracking data, the smaller the predicted gain reward.

[0084] In this embodiment, the better the working conditions and the higher the solar radiation, the lower the training basic reward, but the corresponding prediction gain is higher; the worse the working conditions and the lower the solar radiation, the higher the training basic reward, but the corresponding prediction gain reward is lower. In particular, when it is rainy or snowy, the corresponding training basic reward is equal to the preset reward threshold, and the prediction gain reward is zero, thereby achieving a preset reward threshold to ensure that for each historical working condition, the trained sunshine tracking reinforcement learning model can predict the real-time predicted light tracking data for direct sunlight tracking, thereby providing a basis for accurately adjusting the azimuth servo and altitude servo to enable the solar panels to accurately receive direct sunlight.

[0085] S24. According to the reward of the current round of training, the network parameters of the sunshine tracking reinforcement learning model are adjusted to obtain the initial network parameters for the next round of simulation training;

[0086] S25. Based on S21-S24, several rounds of iterative training are performed until the training rewards corresponding to each historical working condition except rainy and snowy weather are greater than the preset reward threshold, and the training is stopped to obtain a trained sunshine tracking reinforcement learning model.

[0087] S3, according to the historical working condition data of the solar tracking system and the corresponding light tracking data, the control parameters of the PID controller used to control the azimuth servo and the altitude servo are updated to obtain the target PID control parameters;

[0088] The S3 comprises the following steps:

[0089] S31, setting initial control parameters of the PID controller for controlling the azimuth servo and the altitude servo, wherein the control parameters include proportional gain, integral time constant and differential gain; in this embodiment, the proportional gain controls the response speed of the PID controller to the deviation of the deflection of the azimuth servo and the altitude servo, the integral time constant controls the response speed of the controller to the cumulative deviation value of the deflection of the azimuth servo and the altitude servo, and the differential gain controls the response speed of the PID controller to the rate of change of the azimuth servo and the altitude servo;

[0090] S32, according to the historical working condition data of the solar tracking system and the corresponding light tracking data, obtaining the deviation between the dual-axis deflection data under different working conditions and the expected dual-axis deflection data; in this embodiment, the expected dual-axis deflection data is the deflection data of the azimuth servo and the altitude servo when the solar panels in the solar tracking system receive sunlight vertically;

[0091] S33, constructing a control law model and a PID transfer model of the PID controller according to the initial control parameters of the PID controller and the deviation between the dual-axis deflection data under different working conditions and the expected dual-axis deflection data;

[0092] The calculation expressions of the control law model and the PID transfer model of the PID controller in S33 are as follows:

[0093]

[0094] Where u(t) represents the output value of the PID controller driving the azimuth and elevation servos, K p represents the proportional gain, e(t) represents the deviation between the dual-axis deflection data under different working conditions and the expected dual-axis deflection data, T i represents the integral time constant, dt represents the differential with respect to time, It represents the differential of the deviation between the biaxial deflection data under different working conditions and the expected biaxial deflection data with respect to time, T d Denotes the differential gain, K i Indicates the integral coefficient, K d represents the differential coefficient, G(s) represents the transfer function of the PID controller, and s represents the open-loop frequency.

[0095] S34, based on the control law model and PID transfer model of the PID controller, update the control parameters of the PID controller until the deviation between the dual-axis deflection data under different working conditions and the expected dual-axis deflection data is less than a preset deviation threshold, and obtain the target PID control parameters.

[0096] S4. Acquire real-time working condition data of the solar tracking system. In this embodiment, the real-time working condition data of the solar tracking system corresponds to the historical working condition data of the solar tracking system, and both include real-time light data, temperature data, air humidity data, wind speed data, weather data, and power grid data of the solar tracking system.

[0097] S5. According to the real-time working condition data of the solar tracking system, based on the trained sunshine tracking reinforcement learning model and target PID control parameters, the solar tracking system is driven and controlled to perform light tracking.

[0098] The S5 comprises the following steps:

[0099] S51, inputting the real-time working condition data into the trained sunshine tracking reinforcement learning model to obtain the predicted sunshine tracking data corresponding to the real-time working condition;

[0100] S52, if the real-time weather data is rainy or snowy, and the solar panel expansion and contraction data in the predicted light tracking data corresponding to the real-time working condition is that the solar panel is in a contracted state, then the solar panel of the solar tracking system is contracted until the weather data is sunny or cloudy, then the solar panel is expanded, and the process proceeds to S53;

[0101] S53, according to the dual-axis deflection data in the predicted light tracking data corresponding to the real-time working condition, based on the target PID control parameters, use the PID controller to control the driving direction angle servo and the altitude angle servo to deflect, so as to control the solar tracking system to perform light tracking.

[0102] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A driving control method for a solar tracking system, characterized in that: The steps include: S1. Obtain historical operating data of the solar tracking system and light tracking data corresponding to the historical operating data; S2. According to the reinforcement learning method, the historical working condition data of the solar tracking system and the corresponding light tracking data, the light tracking of the solar tracking system is simulated and trained under different working conditions to obtain a trained sunlight tracking reinforcement learning model; S3, according to the historical working condition data of the solar tracking system and the corresponding light tracking data, the control parameters of the PID controller used to control the azimuth servo and the altitude servo are updated to obtain the target PID control parameters; S4, obtaining real-time working data of the solar tracking system; S5. According to the real-time working condition data of the solar tracking system, based on the trained sunshine tracking reinforcement learning model and target PID control parameters, the solar tracking system is driven and controlled to perform light tracking.

2. The driving control method of the solar tracking system according to claim 1, characterized in that: The historical operating data of the solar tracking system in S1 includes light data, temperature data, air humidity data, wind speed data, weather data and power grid data; The illumination data includes the solar altitude angle and the solar azimuth angle; the weather data includes sunny, cloudy, rainy and snowy weather; the power grid data includes frequency and voltage; The light tracking data includes solar panel retraction and expansion data and dual-axis deflection data. The solar panel retraction and expansion data is state data of the solar panel in a retracted state or an expanded state. The dual-axis deflection data includes horizontal angle deflection data and altitude angle deflection data.

3. The driving control method of the solar tracking system according to claim 2, characterized in that: The calculation expressions of the solar altitude angle and azimuth angle are as follows: θ s =(t z -12)×15°,t z =t p +Δ z ,t p =t bj +t xz , D z (rad)=0.00076+0.001869cosθ0+0.014616cos2θ0-0.04085sin2θ0, Among them, Hs represents the solar altitude angle, °S represents the local latitude, σ represents the declination angle, θ s represents the solar hour angle, As represents the solar azimuth, σ (rad) represents the radian corresponding to the declination angle, N represents the number of days that have passed in the year, and t z represents true solar time, t p represents mean solar time, Δ z Indicates the true solar time difference, t bj Indicates Beijing time, t xz represents the accuracy correction time, °E represents the local longitude, Δ z (rad) represents the radian corresponding to the true solar time difference, θ0 represents the day angle, and the solar azimuth is the azimuth of 0° in the south direction.

4. The driving control method of the solar tracking system according to claim 2, characterized in that: The S2 comprises the following steps: S21. According to the reinforcement learning method, the network architecture and initial network parameters of the sunshine tracking reinforcement learning model in the current round of training are set; S22, inputting the historical operating condition data of the solar tracking system into the light tracking reinforcement learning model in sequence for simulation training, and obtaining predicted light tracking data corresponding to the historical operating conditions; S23, calculating the reward for the current round of training based on the reward model according to the predicted light tracking data corresponding to different historical working conditions; S24. According to the reward of the current round of training, the network parameters of the sunshine tracking reinforcement learning model are adjusted to obtain the initial network parameters for the next round of simulation training; S25. Based on S21-S24, several rounds of iterative training are performed until the training rewards corresponding to each historical working condition except rainy and snowy weather are greater than the preset reward threshold, and the training is stopped to obtain a trained sunshine tracking reinforcement learning model.

5. The driving control method of the solar tracking system according to claim 4, characterized in that: The calculation expression of the reward model in S23 is as follows: R cur =(1-ε(t))×R base +ε(t)×R pbr , Among them, R cur represents the reward of the current round of training, ε(t) represents the preset dual-axis tracking time function, R base represents the training base reward, R pbr Represents the prediction gain reward.

6. The driving control method of the solar tracking system according to claim 5, characterized in that: The training basic reward in S23 is related to the illumination data, temperature data, air humidity data, wind speed data, weather data and power grid data during each round of training. Different and fixed training basic rewards correspond to different working conditions. The predicted gain reward is correlated with the difference between the light tracking data corresponding to the historical working condition data and the predicted light tracking data in each round of training, which is reflected in the following two aspects: on the one hand, under the same historical working condition data, the more power the solar panel generates and the smaller the difference between the light tracking data and the predicted light tracking data, the greater the predicted gain reward; on the other hand, under the same historical working condition data, the more power the solar panel generates and the greater the difference between the light tracking data and the predicted light tracking data, the smaller the predicted gain reward.

7. The driving control method of the solar tracking system according to claim 6, characterized in that: The S3 comprises the following steps: S31, setting initial control parameters of a PID controller for controlling the azimuth servo and the altitude servo, wherein the control parameters include a proportional gain, an integral time constant and a differential gain; S32, obtaining the deviation between the dual-axis deflection data under different working conditions and the expected dual-axis deflection data according to the historical working condition data of the solar tracking system and the corresponding light tracking data; S33, constructing a control law model and a PID transfer model of the PID controller according to the initial control parameters of the PID controller and the deviation between the dual-axis deflection data under different working conditions and the expected dual-axis deflection data; S34, based on the control law model and PID transfer model of the PID controller, update the control parameters of the PID controller until the deviation between the dual-axis deflection data under different working conditions and the expected dual-axis deflection data is less than a preset deviation threshold, and obtain the target PID control parameters.

8. The driving control method of the solar tracking system according to claim 7, characterized in that: The calculation expressions of the control law model and the PID transfer model of the PID controller in S33 are as follows: Where u(t) represents the output value of the PID controller driving the azimuth and elevation servos, K p represents the proportional gain, e(t) represents the deviation between the dual-axis deflection data under different working conditions and the expected dual-axis deflection data, T i represents the integral time constant, dt represents the differential with respect to time, It represents the differential of the deviation between the biaxial deflection data under different working conditions and the expected biaxial deflection data with respect to time, T d Denotes the differential gain, K i Indicates the integral coefficient, K d represents the differential coefficient, G(s) represents the transfer function of the PID controller, and s represents the open-loop frequency.

9. The driving control method of the solar tracking system according to claim 7, characterized in that: The S5 comprises the following steps: S51, inputting the real-time working condition data into the trained sunshine tracking reinforcement learning model to obtain the predicted sunshine tracking data corresponding to the real-time working condition; S52, if the real-time weather data is rainy or snowy, and the solar panel expansion and contraction data in the predicted light tracking data corresponding to the real-time working condition is that the solar panel is in a contracted state, then the solar panel of the solar tracking system is contracted until the weather data is sunny or cloudy, then the solar panel is expanded, and the process proceeds to S53; S53, according to the dual-axis deflection data in the predicted light tracking data corresponding to the real-time working condition, based on the target PID control parameters, use the PID controller to control the driving direction angle servo and the altitude angle servo to deflect, so as to control the solar tracking system to perform light tracking.

Images