Commercial power complementation control method of solar lamp controller

Abstract

The invention belongs to the technical field of street lamp controllers, and particularly relates to a mains supply complementation control method of a solar lamp controller, which comprises a street lamp controller, a solar panel electrically connected with the street lamp controller and an LED street lamp, and is characterized in that the street lamp controller is also electrically connected with mains supply through a switching power supply to realize the mains supply power supply action of the LED street lamp. The street lamp controller is further electrically connected with a storage battery used for storing electric energy of the solar cell panel. According to the commercial power complementation control method of the solar lamp controller, the power generation data in each time period in the future can be estimated according to the statistical data, and accurate data is provided for the power supply department of the commercial power how to more accurately deploy the commercial power and the power generation amount. More importantly, the time, the road section and the power of mains supply switching can be allocated according to the generating capacity. Therefore, the problems of current, voltage and the like when all the mains supply street lamps are started at the same time are solved, and the problems of power load and circuit pressure bearing caused by instantaneous mains supply switching are prevented.

Description

Technical Field

[0001] The present invention relates to the technical field of street lamp controllers, and in particular to a mains power complementary control method of a solar lamp controller. Background Art

[0002] In the prior art, for example, a wind-photovoltaic hybrid controller and a photovoltaic charging system with the announcement number CN 108270282 A disclosed on the Chinese patent website can realize the switching of input power through a central microprocessor and an input switching circuit. It only needs to connect the AC input to the AC interface circuit, connect the solar panel to the solar panel interface circuit, and connect the wind turbine to the wind turbine interface circuit, so as to achieve automatic switching and adjustment of AC, photovoltaic and wind power, which is highly convenient.

[0003] The above-mentioned prior art merely solves the problem of how to control the switching between the wind turbine generator and the mains.

[0004] Another example is a wind-solar-electricity complementary street lamp with publication number CN 205504812 U. It mainly generates electricity from solar energy on sunny days, uses wind power when there is wind on rainy days, and uses city electricity to supplement power supply when there is no wind on rainy days. While making full use of natural energy, it also ensures the normality and continuity of road lighting, and has the significant advantage of ensuring that street lamps are lit for a long time and all day long. The wind-solar-electricity complementary controller set in the wind-solar-electricity complementary street lamp displays the street lamp system status and system parameters in intuitive digital and graphical forms, with strong functionality and good economic and technical benefits.

[0005] The above-mentioned prior art also only solves the problem of fully combining photovoltaic power generation on sunny days with wind power generation on rainy days.

[0006] The ideas of the above existing technologies are to solve the coupling and efficiency problems between the mains and photovoltaic or wind power. However, they still cannot solve the balance problem between the entire power generation system and the mains, because this is equivalent to the mains needing to be enabled at any time, which will cause the following problems:

[0007] 1. When the mains power is switched on for a moment, its power will increase with the number of street lamps, which can easily cause power system instability problems such as instantaneous current shock, voltage fluctuation, power supply equipment overload and power factor reduction due to instantaneous start-up of the entire mains power.

[0008] 2. It is difficult to manage the street light system in the whole city in a unified way, that is, each uses its own electricity, and it is unknown how long the photovoltaic power will actually supply power. However, once the city power is switched on, the entire street light system starts up at the same time. This brings great difficulties to the power department of the city power, especially the power generation department, in managing the power generation. Summary of the invention

[0009] Based on the existing technical problems, the present invention proposes a mains power complementary control method for a solar lamp controller.

[0010] The present invention proposes a method for complementary control of a solar lamp controller with a mains supply, comprising a street lamp controller, a solar cell panel electrically connected to the street lamp controller, and an LED street lamp. The street lamp controller is also electrically connected to the mains supply through a switching power supply to realize the mains supply action of the LED street lamp. The street lamp controller is also electrically connected to a battery for storing electric energy of the solar cell panel.

[0011] The method for the street lamp controller to control the solar panel to complement the mains electricity includes:

[0012] Step 1: Collect the past illumination data of the area where the solar panel is installed to form a three-dimensional illumination data set D[n1(x1, y1, z1), n2(x2, y2, z2), n3(x3, y3, z3)...n i (x i ,y i , z i )], where i≥1.

[0013] Step 2: respectively analyze the light intensity data x i , light time data y i And temperature data z i Set an adjustable threshold r(r x , r y , r z ) to form new light intensity data x i ±r x , light time data y i ±r y And temperature data z i ±r z .

[0014] Constructing a new 3D lighting dataset

[0015] The triangular prism-shaped data set composed of its maximum and minimum values ​​is the database of the entire system.

[0016] Step 3: New 3D lighting dataset

[0017] The constructed database is input into the reinforcement learning module in the street light controller for reinforcement learning to obtain the solar panel illumination data point set D that is closest to the actual illumination data point set. 优 [n1(x1,y1,z1),n2(x2,y2,z2),n3(x3,y3,z3)......ni (x i ,y i , z i )], and then these illumination data sets D 优 [n1(x1,y1,z1),n2(x2,y2,z2),n3(x3,y3,z3)......n i (x i ,y i , z i )] are connected by a straight line L to obtain the optimal power generation illumination data function f[n 优 (x 优 ,y 优 , z 优 )], thus obtaining the theoretical total power generation C of the solar panel 理论 .

[0018] Step 4: According to the theoretical total power generation C of the battery in step 2 理论 , multiplied by the loss coefficient k, where k≤1, to obtain the actual total power consumption C 实际 =C 理论 *k;

[0019] According to the actual total power consumption C 实际 And the total power consumption W of the LED street lamp 总功率 Calculate the actual lighting time of the LED street lamp Finally, the controller issues an instruction on whether to connect to the mains power.

[0020] Step 5: According to the power generation efficiency of the solar panel, the three-dimensional illumination data set D[n1(x1, y1, z1), n2(x2, y2, z2), n3(x3, y3, z3)...n i (x i ,y i , z i )] to perform data statistics and calculations, and finally determine whether the solar cell panel is damaged.

[0021] Preferably, in step 1, a data set consisting of several key data affecting power generation efficiency is collected and analyzed, and the data set includes the illumination data points n i (x i ,y i , z i ) light intensity data x i , light time data y i And temperature data z i , the illumination data point n i (x i ,y i , zi ) with adjustable interval time t i Conduct collection;

[0022] Then, taking the origin n0 (x0, y0, z0) of the three-dimensional coordinate system as the starting point, the three-dimensional illumination data set D[n1 (x1, y1, z1), n2 (x2, y2, z2), n3 (x3, y3, z3) ... n i (x i ,y i , z i )] for each illumination data point n i (x i ,y i , z i ) are connected in sequence with straight lines.

[0023] Preferably, in step 2, since the units of the light intensity, light exposure time and temperature are different, it is necessary to set the threshold value r separately for each of them. x , threshold r y and the threshold r z , and the light intensity data x i , light time data y i And temperature data z i Attach the threshold r x 、r y and r z After that, the maximum and minimum values ​​of each data will be formed, and then the new light intensity data x will be formed. i ±r x , light time data y i ±r y And temperature data z i ±r z .

[0024] Preferably, the light intensity data x i ±r x , light time data y i ±r y And temperature data z i ±r z After attaching the respective thresholds, the difference between the maximum and minimum values ​​is doubled, so is the real data set with the added threshold, and these new 3D illumination data sets

[0025] The triangular prism-shaped data set composed of its maximum and minimum values ​​is the database of the entire system.

[0026] Preferably, in step 3, the database serves as the environment of the agent, forming a complete loop of reinforcement learning: the street lamp controller agent in the learning state s t Perform action a t , and get rewards from the database environment t , arrive at state s t+1 ; where t is the number of time steps in the future where the reward will be.

[0027] Preferably, in the agent, the reward r is not obtained at each step. t And rewards t The size of is the same, so we need to add an additional parameter 0≤γ≤1, and then when the kth reward comes, the total reward R t for:

[0028]

[0029] Preferably, for the street lamp controller agent, there are many paths from the starting data point to the end data point, and there are many learning paths from the first step to the second step and the rewards obtained, so it is necessary to try to decide which action to choose to maximize the expected reward in the future. For data, this expected reward is the value, and this value is calculated by selecting the action value function Q(s, a):

[0030]

[0031] The best strategy for lighting data π * For all possible states s and actions a, the optimal action value function is Then select action a, according to the optimal strategy π * Make a choice, so the current reward r t and the estimate of the (discounted) future rewards is:

[0032]

[0033] Q-learning enables the street light controller agent to choose the best action in the environment by learning the optimal strategy. The calculation formula is:

[0034]

[0035] Among them, s is the current state; a is the current action; r is the reward of the current action; s′ is the next state; α is the learning rate; γ is the discount factor; is the action with the maximum Q value among all possible actions in the next state.

[0036] Preferably, the illumination data function The corners of the line connecting each data point are smoothly transitioned using an arc line L(r′) protruding outward from the intersection of two straight lines, where the arc radius r′ of the transition arc line L(r′) ranges from in, represents the shortest half length of two adjacent straight lines, and Less than or equal to the threshold r x 、r y and r z The smallest threshold value.

[0037] Preferably, in step 5, each of the solar panels is numbered m1, m2...m j ; Wherein j≥1, all of the solar panels are connected to the street light controller, and their numbers are marked one by one in the street light controller.

[0038] Preferably, a single solar cell panel m is installed in the street light controller. j Then number them into individual power generation modules: m′1, m′2...m′ g ; Among them, g ≥ 1;

[0039] When the single solar panel m is below the threshold j All individual power generation modules m′ g When the power generation is within the range of threshold 2, it proves that all single power generation modules m′ g are covered by dust, resulting in the power generation of the entire panel being lower than the threshold value 1. When one or more of the individual power generation modules m′ g When the power generation is lower than the threshold value 2, it proves that this one or several single power generation modules m′ g Covered by the cover, then according to this one or several single power generation modules m' g The position of the covering object can be known.

[0040] The beneficial effects of the present invention are:

[0041] 1. Collect historical solar energy data at the site where solar panels are installed to estimate whether solar panels are economically worthwhile to build.

[0042] 2. Based on the statistical data, the power generation data in each time period in the future can be estimated, and the power supply department of the city power can be provided with more accurate data on how to more accurately allocate the city power and power generation. More importantly, the time, section, and power of the city power switching can be allocated according to the power generation. This can reduce the current and voltage problems when all the city power lights are started at the same time, and prevent the occurrence of power load and line pressure problems caused by instantaneous switching of the city power.

[0043] 3. The power generation data of each solar panel can also be used to determine whether there are foreign objects on the solar panel, whether it is damaged, and whether staff need to be dispatched to clean it. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 A schematic diagram of a mains-electricity complementary control method for a solar lamp controller proposed by the present invention;

[0045] Figure 2 A physical connection diagram of a mains complementary control method for a solar lamp controller proposed by the present invention;

[0046] Figure 3 The illumination data point n of the mains complementary control method of a solar lamp controller proposed by the present invention i (x i ,y i , z i ) three-dimensional distribution map;

[0047] Figure 4 The illumination data point n of the mains complementary control method of a solar lamp controller proposed by the present invention i (x i ,y i , z i )Set the threshold r(r x , r y , r z ) after the schematic diagram;

[0048] Figure 5 Data set D for the mains complementary control method of a solar lamp controller proposed by the present invention r Schematic diagram of the constructed database;

[0049] Figure 6 A complete circuit diagram of a reinforcement learning module of a mains complementary control method for a solar lamp controller proposed by the present invention;

[0050] Figure 7 A schematic diagram of reinforcement learning of a reinforcement learning module of a mains complementary control method of a solar lamp controller proposed by the present invention;

[0051] Figure 8 The optimal power generation illumination data function f[n] after reinforcement learning of the reinforcement learning module of the mains complementary control method of the solar lamp controller proposed by the present invention 优 (x 优 ,y 优 , z 优 )] Schematic diagram;

[0052] Fig. 9A schematic diagram of the smooth transition of the arc line L(r′) of the mains complementary control method of a solar lamp controller proposed by the present invention;

[0053] Fig.10 A solar panel m for a city power complementary control method of a solar lamp controller proposed by the present invention j Numbering diagram;

[0054] Fig.11 A single power generation module m' of a city power complementary control method for a solar lamp controller proposed by the present invention g Numbering diagram. DETAILED DESCRIPTION

[0055] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0056] Reference Figure 1-Figure 11 , a method for complementary control of a solar lamp controller with a mains supply, such as Figure 1 As shown, it includes a street light controller, a solar cell panel electrically connected to the street light controller, and an LED street light. The street light controller is also electrically connected to the mains via a switching power supply to realize the mains power supply of the LED street light. The street light controller is also electrically connected to a battery for storing electrical energy from the solar cell panel.

[0057] The so-called mains-electricity complementary LED solar street light system is an LED street light lighting system that mainly uses solar cells to generate electricity and supplements it with ordinary 220V AC electricity. With this system, the photovoltaic battery group and battery capacity can be designed to be smaller. Basically, when there is sunshine during the day, solar power is used to charge the battery, and when it gets dark, the battery discharges to light up the load LED. However, in some areas, there are basically more than 2 / 3 sunny days throughout the year. In this way, the system uses solar energy to illuminate the street lights for more than 2 / 3 of the year, and uses mains electricity to supplement energy for the rest of the time. This not only reduces the one-time investment of the solar photovoltaic lighting system, but also has a significant energy-saving and emission-reduction effect. It is an effective method for promoting and popularizing solar LED street lighting at this stage.

[0058] 3.2V system mains complementary solar lights such as Figure 2 As shown, it consists of five parts: solar panels, street light controller, battery pack, switching power supply and LED light source.

[0059] Among them, the street light controller is the control core of the whole system, which is used for controlling the charging and discharging of the battery, the lighting of the LED street lights, the automatic switching control of the mains power supply and related protection control; the solar panel is used to convert solar energy into electrical energy, and the street light controller determines whether it is daytime or nighttime by collecting the size of the solar energy output voltage to control the turning on and off of the light; the battery is the energy storage device of the whole system, the solar panel charges the battery during the day, and the LED street lights are powered by the battery at night or on rainy days; the switching power supply converts the 220V mains power into DC12V 5A direct current, which is used to power the light source through the street light controller when the battery is low.

[0060] The working principle of this system is that the solar panel charges the battery during the day, and the LED light source is powered by the battery at night or on rainy days. The controller collects the battery voltage in real time, and automatically switches to the mains power supply to ensure the brightness of the light when the battery power is insufficient (the voltage is lower than 3.0V). Because of the support of the mains, the mains complementary system is more stable and not affected by the weather compared to the pure solar energy system, but because it needs to lay cables with the mains, it completely loses the advantage of solar street lights being easy to install.

[0061] In order to achieve the complementary effect between solar power generation and mains power supply, the method of controlling the solar panel to complement the mains power by the street lamp controller includes:

[0062] Step 1: Figure 3 As shown, the past illumination data of the area where the solar panel is installed is collected to form a three-dimensional illumination data set D[n1(x1, y1, z1), n2(x2, y2, z2), n3(x3, y3, z3)......n i (x i ,y i , z i )], where i≥1.

[0063] Since the illumination data directly affects the power generation efficiency of the solar panel, firstly, a data set consisting of several key data affecting the power generation efficiency is collected and analyzed. This data set includes the illumination data points n i (x i ,y i , z i ) light intensity data x i , light time data y i And temperature data z i The illumination data point n i (x i ,y i , z i ) with adjustable interval time t i Collect.

[0064] Then, taking the origin of the three-dimensional coordinate system n0 (x0, y0, z0) as the starting point, the three-dimensional illumination data set D[n1 (x1, y1, z1), n2 (x2, y2, z2), n3 (x3, y3, z3) ... n i (x i ,y i , z i )] for each illumination data point n i (x i ,y i , z i ) are connected in sequence with straight lines.

[0065] Step 2: Figure 4 As shown, the light intensity data x i , light time data y i And temperature data z i Set an adjustable threshold r(r x , r y , r z ) to form new light intensity data x i ±r x , light time data y i ±r y And temperature data z i ±r z Because the units of light intensity, light duration and temperature are different, the threshold r needs to be set separately for each of them. x , threshold r y and the threshold r z , and the light intensity data x i , light time data y i And temperature data z i Attach the threshold r x 、r y and r z After that, the maximum and minimum values ​​of each data will be formed, and then the new light intensity data x will be formed. i ±r x , light time data y i ±r y And temperature data z i ±r z , thus forming a new three-dimensional lighting dataset

[0066] Finally, these data points are connected in sequence with straight lines to form a prismatic data set. The data set at any time point is the cross section of the cylindrical data set, forming a triangle. x 、r yand r z It is the difference between the maximum value and the minimum value, so the light intensity data x i ±r x , light time data y i ±r y And temperature data z i ±r z After attaching the respective thresholds, the difference between the maximum and minimum values ​​is doubled, so is the real data set with the threshold added, such as Figure 5 As shown, these new 3D lighting datasets

[0067] The triangular prism-shaped data set composed of its maximum and minimum values ​​is the database of the entire system.

[0068] Step 3: Figure 3-Figure 5 As shown, the new 3D lighting dataset

[0069] The constructed database is input into the reinforcement learning module in the street light controller for reinforcement learning to obtain the solar panel illumination data point set D that is closest to the actual illumination data point set. 优 [n1(x1,y1,z1),n2(x2,y2,z2),n3(x3,y3,z3)......n i (x i ,y i , z i )], and then these illumination data sets D 优 [n1(x1,y1,z1),n2(x2,y2,z2),n3(x3,y3,z3)......n i (x i ,y i , z i )] are connected by straight line L to obtain the optimal power generation illumination data function Thus, the theoretical total power generation C of the solar panel is obtained. 理论 .

[0070] The reinforcement learning module is implemented as follows:

[0071] like Figure 6-Figure 8 As shown, the reinforcement learning module is a mapping between states or situations and actions, the goal is to maximize some numerical rewards. In the database formed, if the street light controller wants to actively find one or more D 优[n1(x1,y1,z1),n2(x2,y2,z2),n3(x3,y3,z3)......n i (x i ,y i , z i )] constitutes the optimal power generation illumination data function f[n 优 (x 优 ,y 优 , z 优 )], reinforcement learning is needed to achieve: the street light controller is used as an intelligent agent, and a new 3D lighting dataset The database is used as the environment of the agent, which can form a complete loop of reinforcement learning: the learned street light controller agent is in state s t Perform action a t , and get rewards from the database environment t , arrive at state s t+1 ; where t is the number of time steps in the future where the reward will be.

[0072] Since in the agent, not every step gets a reward r t And rewards t The size of is the same, so we need to add an additional parameter 0≤γ≤1, and then when the kth reward comes, the total reward R t for:

[0073]

[0074] For the street light controller agent, there are many paths from the starting data point to the end data point, and there are many learning paths from the first step to the second step and the rewards obtained, so it is necessary to try to decide which action to choose to maximize the expected reward in the future. For the data, this expected reward is the value. There are two ways to calculate this value: one is to consider the current state and average all possible actions, letting the strategy solve the problem itself. This method is called the state value function V(s); the other is to consider the current state and each possible action separately. This is called the action value function Q(s, a). Due to the new three-dimensional lighting dataset

[0075] The database has enough local illumination data samples from previous years, and the illumination data has been thresholded r x 、r y and r z At the same time, in order to more accurately calculate the illumination data in a certain period of time in the future after the illumination data reinforcement learning, the action value function Q(s, a) is selected in the above two calculation methods:

[0076]

[0077] At this time, the best strategy π for illumination data * For all possible states s and actions a, the optimal action value function is Then select action a according to the optimal strategy π * Make a choice, so the current reward r t and the estimate of the (discounted) future rewards is:

[0078]

[0079] The current reward r t and the (discounted) future rewards are only estimates because the system in the street light controller may have errors and the 3D illumination database D r Medium illumination data point n i (x i ,y i , z i ) will also limit the exploration performed by the agent, so it is sufficient for the street light controller system to learn a good approximation of the optimal strategy.

[0080] Because the data in the database of this system collects illumination data, and when a year is a cycle, the illumination and temperature are similar in adjacent consecutive days, that is to say, the illumination data in a year is actually relatively concentrated and gentle. Except for the half month before and after the four-level alternation, there will be large changes in illumination and temperature data. The changes in illumination data at other times are relatively concentrated and gentle. And because the future data is predicted based on the data of previous years, the offline reinforcement learning algorithm will be more in line with the actual situation. Therefore, it is more appropriate to use the Q-learning algorithm. Q-learning is an offline learning algorithm that aims to learn a state-action value function (Q function), which can be used to evaluate the quality of taking an action in a given state. Maximize the expected cumulative reward for achieving the goal. Q-learning enables the street light controller agent to choose the optimal action in the environment by learning the optimal strategy. Its calculation formula is:

[0081]

[0082] Among them, s is the current state; a is the current action; r is the reward of the current action; s′ is the next state; α is the learning rate; γ is the discount factor; is the action with the maximum Q value among all possible actions in the next state.

[0083] From then on, reinforcement learning of lighting data is completed.

[0084] Finally, if Fig. 9 As shown, the data points are connected by a straight line L to obtain the optimal power generation illumination data function Through the optimal power generation light data function The displayed data can predict or calculate the optimal power generation data for a certain period of time in the future. Combined with the power generation data of previous years, it can be known how much city electricity is needed in a certain period of time in the future and the corresponding time point for city electricity access.

[0085] In order to make the optimal power generation illumination data function The function line is continuous and smooth, and the illumination data function The corners of the line connecting each data point are smoothly transitioned using an arc line L(r′) protruding outward from the intersection of two straight lines, where the arc radius r′ of the transition arc line L(r′) ranges from in, represents the shortest half length of two adjacent straight lines, and Less than or equal to the threshold r x 、r y and r z The minimum threshold is set in order to prevent the arc line L(r′) from exceeding the threshold and affecting the calculation of power generation efficiency.

[0086] In order to calculate the actual power generation and actual power consumption more accurately, set step 4 to calculate the actual total power generation C 实际 ; According to the theoretical total power generation of the battery C 理论 , multiplied by the loss coefficient k, the actual total power consumption C is obtained 实际 =C 理论 *k. According to the actual total electricity consumption C 实际 And the total power consumption W of the LED street lamp 总功率 Calculate the actual lighting time of the LED street lamp Finally, the controller issues an instruction on whether to connect to the mains and stores the power in the battery for standby use. Where k≤1, the loss coefficient includes the loss in the power generation conversion process, the loss in the battery discharge process and the minimum power of the battery.

[0087] In order to calculate the power generation data more accurately, step 5 is set to calculate the three-dimensional illumination data set D[n1(x1, y1, z1), n2(x2, y2, z2), n3(x3, y3, z3)......n i (x i ,y i , z i)] to perform data statistics and calculations, and finally determine whether the solar panel is damaged. The specific statistics and calculations are achieved as follows:

[0088] like Fig.10 As shown, each of the solar panels is numbered m1, m2...m j ; Wherein j≥1. All the solar panels are connected to the street light controller, and their numbers are marked one by one in the street light controller.

[0089] Afterwards Fig.11 As shown, when there is foreign matter on one or more of the solar panels, its power generation will be counted by the street light controller and then compared with other panels with normal power generation. When the comparison result is lower than the set threshold value, it can be known whether there is foreign matter on the current solar panel.

[0090] In order to more accurately determine whether the foreign object is long-accumulated dust or a cover, a single solar panel m is placed in the street light controller. j Then number them into individual power generation modules: m′1, m′2...m′ g ; where g≥1.

[0091] When the single solar panel m is below the threshold j All individual power generation modules m′ g When the power generation is within the range of threshold 2, it proves that all single power generation modules m′ g are covered by dust, resulting in the power generation of the entire panel being lower than threshold 1. g When the power generation is lower than the threshold value 2, it proves that this one or several single power generation modules m′ g Covered by the cover, then according to this one or several single power generation modules j z The position of the covering object can be known.

[0092] It is also possible to determine whether the covering is a light floating object such as leaves by comparing the power generation data in the next few days. If it is a floating object, it will be blown away by the wind in the next few days and power generation will return to normal. g , it will not be blown away by the wind, so it can be determined that the covering is pollutants such as bird droppings, reminding the staff to clean it up in time.

[0093] By implementing the above steps, the following effects can be achieved:

[0094] 1. Collect historical solar energy data at the site where solar panels are installed to estimate whether solar panels are economically worthwhile to build.

[0095] 2. Based on the statistical data, the power generation data in each time period in the future can be estimated, and the power supply department of the city power can be provided with more accurate data on how to more accurately allocate the city power and power generation. More importantly, the time, section, and power of the city power switching can be allocated according to the power generation. This can reduce the current and voltage problems when all the city power lights are started at the same time, and prevent the occurrence of power load and line pressure problems caused by instantaneous switching of the city power.

[0096] 3. The power generation data of each solar panel can also be used to determine whether there are foreign objects on the solar panel, whether it is damaged, and whether staff need to be dispatched to clean it.

[0097] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A method for controlling a solar lamp controller with a mains supply, comprising a street lamp controller, a solar panel electrically connected to the street lamp controller, and an LED street lamp, wherein the street lamp controller is also electrically connected to the mains supply through a switching power supply to realize the mains supply of the LED street lamp, and the street lamp controller is also electrically connected to a battery for storing the electric energy of the solar panel; Features: The method for the street lamp controller to control the solar panel to complement the mains electricity includes: Step 1: Collect the past illumination data of the area where the solar panel is installed to form a three-dimensional illumination data set D[n1(x1, y1, z1), n2(x2, y2, z2), n3(x3, y3, z3)...n i (x i ,y i , z i )], where i≥1; Step 2: respectively analyze the light intensity data x i , light time data y i And temperature data z i Set an adjustable threshold r(r x , r y , r z ) to form new light intensity data x i ±r x , light time data y i ±r y And temperature data z i ±r z ; Constructing a new 3D lighting dataset The triangular prism-shaped data set composed of its maximum and minimum values ​​is the database of the entire system; Step 3: New 3D lighting dataset The constructed database is input into the reinforcement learning module in the street light controller for reinforcement learning to obtain the solar panel illumination data point set D that is closest to the actual illumination data point set. 优 [n1(x1,y1,z1),n2(x2,y2,z2),n3(x3,y3,z3)......n i (x i ,y i , z i )], and then these illumination data sets D 优 [n1(x1,y1,z1),n2(x2,y2,z2),n3(x3,y3,z3)......n i (x i ,y i , z i )] are connected by a straight line L to obtain the optimal power generation illumination data function f[n 优 (x 优 ,y 优 , z 优 )], thus obtaining the theoretical total power generation C of the solar panel 理论 ; Step 4: According to the theoretical total power generation C of the battery in step 2 理论 , multiplied by the loss coefficient k, where k≤1, to obtain the actual total power consumption C 实际 =C 理论 *k; According to the actual total power consumption C 实际 And the total power consumption W of the LED street lamp 总功率 Calculate the actual lighting time of the LED street lamp Finally, the controller issues an instruction on whether to connect to the mains power; Step 5: According to the power generation efficiency of the solar panel, the three-dimensional illumination data set D[n1(x1, y1, z1), n2(x2, y2, z2), n3(x3, y3, z3)...n i (x i ,y i , z i )] to perform data statistics and calculations, and finally determine whether the solar cell panel is damaged.

2. The method for controlling the mains electricity of a solar lamp controller according to claim 1, characterized in that: In the step 1, a data set consisting of several key data affecting power generation efficiency is collected and analyzed. This data set includes the illumination data points n i (x i ,y i , z i ) light intensity data x i , light time data y i And temperature data z i , the illumination data point n i (x i ,y i , z i ) with adjustable interval time t i Conduct collection; Then, taking the origin n0 (x0, y0, z0) of the three-dimensional coordinate system as the starting point, the three-dimensional illumination data set D[n1 (x1, y1, z1), n2 (x2, y2, z2), n3 (x3, y3, z3) ... n i (x i ,y i , z i )] for each illumination data point n i (x i ,y i , z i ) are connected in sequence with straight lines.

3. The method for controlling the mains electricity of a solar lamp controller according to claim 2, characterized in that: In step 2, since the units of the light intensity, light exposure time and temperature are different, it is necessary to set the threshold value r separately for each of them. x , threshold r y and the threshold r z , and the light intensity data x i , light time data y i And temperature data z i Attach the threshold r x 、r y and r z After that, the maximum and minimum values ​​of each data will be formed, and then the new light intensity data x will be formed. i ±r x , light time data y i ±r y And temperature data z i ±r z .

4. The method for controlling the solar lamp controller with AC power according to claim 3, characterized in that: The light intensity data x i ±r x , light time data y i ±r y And temperature data z i ±r z After attaching the respective thresholds, the difference between the maximum and minimum values ​​is doubled, so is the real data set with the added threshold, and these new 3D illumination data sets The triangular prism-shaped data set composed of its maximum and minimum values ​​is the database of the entire system.

5. The method for controlling the solar lamp controller with AC power according to claim 4, characterized in that: In step 3, the database serves as the environment of the agent, forming a complete loop of reinforcement learning: the learned street light controller agent is in state s t Perform action a t , and get rewards from the database environment t , arrive at state s t+1 ; where t is the number of time steps in the future where the reward will be.

6. The method for controlling the solar lamp controller with AC power according to claim 5, characterized in that: In the agent, not every step gets a reward r t And rewards t The size of is the same, so we need to add an additional parameter 0≤γ≤1, and then when the kth reward comes, the total reward R t for:

7. The method for controlling the solar lamp controller with AC power according to claim 6, characterized in that: For the street light controller agent, there are multiple paths from the starting data point to the end data point, which means that the learning path from the first step to the second step and the reward obtained are also different, so we need to try to decide which action to choose to maximize the expected reward in the future. For data, this expected reward is the value, and this value is calculated by selecting the action value function Q(s, a): The best strategy for lighting data π * For all states s and actions a, the optimal action value function is Then select action a, according to the optimal strategy π * Make a choice, so the current reward r t And future rewards are: Q-learning enables the street light controller agent to choose the best action in the environment by learning the optimal strategy. The calculation formula is: Among them, s is the current state; a is the current action; r is the reward of the current action; s′ is the next state; α is the learning rate; γ is the discount factor; is the action with the maximum Q value among all possible actions in the next state.

8. The method for controlling the solar lamp controller with AC power according to claim 7, characterized in that: The illumination data function f[n 优 (x 优 ,y 优 , z 优 )] The corners of the line connecting each data point are smoothly transitioned using an arc line L(r′) protruding outward from the intersection of two straight lines, where the arc radius r′ of the transition arc line L(r′) ranges from in, represents the shortest half length of two adjacent straight lines, and Less than or equal to the threshold r x 、r y and r z The smallest threshold value.

9. The method for controlling a solar lamp controller with a mains supply according to claim 1, characterized in that: In step 5, each solar cell panel is numbered m1, m2...m j ; Wherein j≥1, all of the solar panels are connected to the street light controller, and their numbers are marked one by one in the street light controller.

10. A method for controlling a solar lamp controller with a mains supply according to claim 9, characterized in that: The street light controller has a single solar panel m j Then number them into individual power generation modules: m′1, m′2...m′ g ; Among them, g ≥ 1; When the single solar panel m is below the threshold j All individual power generation modules m′ g When the power generation is within the range of threshold 2, it proves that all single power generation modules m′ g are covered by dust, resulting in the power generation of the entire panel being lower than the threshold value 1. When one or more of the individual power generation modules m′ g When the power generation is lower than the threshold value 2, it proves that this one or several single power generation modules m′ g Covered by the cover, then according to this one or several single power generation modules j z The position of the covering object can be known.

Images