Forage seed screening method and device suitable for photovoltaic area

Abstract

The invention provides a grass seed screening method and device suitable for a photovoltaic area, and relates to the technical field of intelligent seed screening, and the method comprises the steps: obtaining historical environment parameters of a target photovoltaic area at each sampling moment in a specified historical time period, and outputting predicted environment parameters of each sampling moment in a specified future time period through an environment prediction model; acquiring actual environment parameters of each to-be-monitored area, determining a sampling moment matched with the current moment in a specified future time period as a target moment, and controlling an environment adjusting device to adjust the environment parameters of each to-be-monitored area based on a matching result of the actual environment parameters of each to-be-monitored area and the predicted environment parameters of the target moment; pasture images of all the to-be-monitored areas are obtained, pasture growth state data are determined, and the seed category corresponding to the pasture growth state data meeting the preset condition is determined as the target seed category. The prediction accuracy of the seed growth environment and the screening accuracy and efficiency of the seeds are effectively improved.

Description

Technical Field

[0001] The present application relates to the technical field of intelligent seed screening, and particularly to a method for screening forage seeds suitable for photovoltaic areas and a device for screening forage seeds suitable for photovoltaic areas. Background Art

[0002] With the continuous advancement of China's clean energy strategy, solar photovoltaic power generation, as an important form of renewable energy, has been widely applied and developed in the northwestern region. However, due to reasons such as solar panel shading, the microclimate conditions of the area have been changed, affecting the natural growth environment of the vegetation directly below the solar panels, resulting in difficult normal growth of the vegetation in this area. For example, the solar panels block direct sunlight, reducing the light intensity received by the ground surface; the photovoltaic modules absorb and convert solar radiant energy, causing changes in the surrounding temperature distribution and possibly forming local temperature differences; in addition, the solar panels may also affect microclimate factors such as wind speed and humidity. Under the combined action of the above influencing factors, the evaporation of soil moisture below the solar panels may be reduced, and the soil temperature may be lowered, thereby inhibiting the growth of vegetation. Currently, the vegetation planting in the photovoltaic panel covered area usually determines the types of seeds to be sown based on experience. However, most seeds are difficult to adapt to the growth environment of this area, often resulting in unsatisfactory vegetation growth in the photovoltaic panel covered area. Summary of the Invention

[0003] The purpose of the embodiments of the present application is to provide a method for screening forage seeds suitable for photovoltaic areas and a device for screening forage seeds suitable for photovoltaic areas to solve the above problems.

[0004] To achieve the above purpose, the first aspect of the present application provides a method for screening forage seeds suitable for photovoltaic areas, including:

[0005] Obtaining historical environmental parameters of the target photovoltaic area at each sampling moment within a specified historical period, using the historical environmental parameters as inputs, and the pre-trained environmental prediction model outputs the predicted environmental parameters of the target photovoltaic area at each sampling moment within a specified future period. The environmental prediction model is obtained by training a preset neural network with the historical environmental parameters of the target photovoltaic area and an improved optimization algorithm. The target photovoltaic area is the shaded area of the solar panel;

[0006] Obtaining the actual environmental parameters of each area to be monitored within the experimental area, determining the sampling moment within the specified future period that matches the current moment as the target moment, and matching the actual environmental parameters of each area to be monitored with the predicted environmental parameters at the target moment to obtain the environmental parameter matching results of each area to be monitored. Among them, multiple forage seeds are pre-sown in each area to be monitored, and the forage seeds within each area to be monitored belong to the same seed type. The forage seeds in all areas to be monitored belong to at least two seed types;

[0007] Based on the environmental parameter matching results of each area to be monitored, control the environmental adjustment device to adjust the environmental parameters of each area to be monitored;

[0008] Obtain the pasture images of each area to be monitored through an image acquisition device, determine the pasture growth state data of each area to be monitored based on the pasture images of each area to be monitored, and determine that the seed category corresponding to the pasture growth state data that meets the preset conditions among all the obtained pasture growth state data is the target seed category.

[0009] Optionally, the environmental parameters include:

[0010] Light intensity, environmental temperature and environmental humidity;

[0011] The preset neural network is a BP neural network, a convolutional neural network or a recurrent neural network;

[0012] The environmental adjustment device includes:

[0013] A lighting device for adjusting the light intensity of the experimental area, a temperature adjustment device for adjusting the environmental temperature of the experimental area, and a humidity adjustment device for adjusting the environmental humidity of the experimental area;

[0014] The image acquisition device includes:

[0015] At least one fixed bracket, a moving component and a multispectral camera;

[0016] The moving component includes a first slide rail, a second slide rail and a third slide rail;

[0017] One end of the fixed bracket is fixed on the ground of the experimental area, and the other end of the fixed bracket is respectively fixedly connected to the first slide rail and the second slide rail. The first slide rail and the second slide rail are arranged in parallel, and the plane where the first slide rail and the second slide rail are located is parallel to the ground;

[0018] The third slide rail is slidably connected to the first slide rail and the second slide rail, and the third slide rail is perpendicular to the first slide rail and the second slide rail. The multispectral camera is slidably connected to the third slide rail.

[0019] Optionally, based on the environmental parameter matching results of each area to be monitored, controlling the environmental adjustment device to adjust the environmental parameters of each area to be monitored includes:

[0020] Taking the actual light intensity of the experimental area as the actual light intensity of each area to be monitored, if the difference between the actual light intensity and the predicted light intensity of each area to be monitored is greater than the light intensity difference threshold, control the lighting device to adjust the actual light intensity of the experimental area until the difference between the actual light intensity and the predicted light intensity of the experimental area is not greater than the light intensity difference threshold;

[0021] Taking the actual ambient temperature of the experimental area as the actual ambient temperature of each area to be monitored, if the difference between the actual ambient temperature and the predicted ambient temperature of each area to be monitored is greater than the ambient temperature difference threshold, control the temperature adjustment device to adjust the actual ambient temperature of the experimental area until the difference between the actual ambient temperature and the predicted ambient temperature of the experimental area is not greater than the ambient temperature difference threshold;

[0022] Taking the actual ambient humidity of the experimental area as the actual ambient humidity of each area to be monitored, if the difference between the actual ambient humidity and the predicted ambient humidity of each area to be monitored is greater than the ambient humidity difference threshold, control the humidity adjustment device to adjust the ambient humidity of the experimental area until the difference between the ambient humidity and the predicted ambient humidity of the experimental area is not greater than the ambient humidity difference threshold.

[0023] Optionally, the environmental parameters further include soil pH value and soil conductivity; the environmental adjustment device further includes:

[0024] A pH adjustment device for adjusting the soil pH value of the corresponding area to be monitored and a soil conductivity adjustment device for adjusting the soil conductivity of the corresponding area to be monitored, wherein the pH adjustment device and the soil conductivity adjustment device correspond to each area to be monitored one by one;

[0025] Based on the environmental parameter matching results of each area to be monitored, controlling the environmental adjustment device to adjust the environmental parameters of each area to be monitored further includes:

[0026] If the difference between the actual soil pH value and the preset soil pH reference value of each area to be monitored is greater than the soil pH difference threshold, control the pH adjustment device to adjust the soil pH value of each area to be monitored until the difference between the actual soil pH value and the soil pH reference value of each area to be monitored is not greater than the soil pH difference threshold;

[0027] If the difference between the actual soil conductivity and the preset soil conductivity reference value of each area to be monitored is greater than the soil conductivity difference threshold, control the soil conductivity adjustment device to adjust the soil conductivity of each area to be monitored until the difference between the actual soil conductivity and the soil conductivity reference value of each area to be monitored is not greater than the soil conductivity difference threshold.

[0028] Optionally, the forage growth status data includes:

[0029] The chlorophyll content, leaf area index, and vegetation index of the forage;

[0030] After determining the forage growth status data of each area to be monitored, the method further includes:

[0031] Sort the chlorophyll content, leaf area index, and vegetation index of the forage corresponding to all areas to be monitored from high to low respectively to obtain a chlorophyll content queue, a leaf area index queue, and a vegetation index queue of the forage;

[0032] The preset conditions include:

[0033] The chlorophyll content of the forage in the currently monitored area is in the top n% of the chlorophyll content queue of the forage, the leaf area index of the currently monitored area is in the top n% of the leaf area index queue, and the vegetation index of the currently monitored area is in the top n% of the vegetation index queue.

[0034] Optionally, after determining the forage growth status data of each area to be monitored, the method further includes:

[0035] Sort the chlorophyll content, leaf area index, and vegetation index of the forage corresponding to all areas to be monitored from high to low respectively to obtain a chlorophyll content queue, a leaf area index queue, and a vegetation index queue of the forage;

[0036] Determine the first weight of the chlorophyll content of the forage, the second weight of the leaf area index, and the third weight of the vegetation index;

[0037] Determine the first score of the chlorophyll content of the forage in the currently monitored area according to the preset forage growth status data scoring table and the position of the chlorophyll content of the forage in the currently monitored area in the chlorophyll content queue of the forage, determine the second score of the leaf area index of the forage in the currently monitored area according to the forage growth status data scoring table and the position of the leaf area index of the forage in the currently monitored area in the leaf area index queue, and determine the third score of the vegetation index of the forage in the currently monitored area according to the forage growth status data scoring table and the position of the vegetation index of the forage in the currently monitored area in the vegetation index queue;

[0038] Perform weighted summation on the first score, the second score, and the third score based on the first weight, the second weight, and the third weight to obtain the forage growth status data scores of each area to be monitored, sort the forage growth status data scores of each area to be monitored from high to low to obtain a forage growth status data score queue;

[0039] The forage growth status data scoring table includes at least a first score corresponding to different positions of the chlorophyll content of the forage in the chlorophyll content queue of the forage, a second score corresponding to different positions of the leaf area index in the leaf area index queue, and a third score corresponding to different positions of the vegetation index in the vegetation index queue;

[0040] The preset conditions include:

[0041] The scoring of the forage growth status data in the current area to be monitored is in the top n% of the forage growth status data scoring queue.

[0042] Optionally, the improved optimization algorithm includes:

[0043] S10. Initialize the particle swarm, determine the number of particles in the particle swarm and the initial positions of each particle, and map the hyperparameters of the preset neural network to different dimensions and positions of each particle;

[0044] S20. Determine the fitness value of each particle based on a preset fitness function, and determine the minimum fitness value, the maximum fitness value, the average fitness value of all current particles, the individual extreme value of each particle, and the group extreme value. The fitness function is constructed based on the output error of the preset neural network;

[0045] S30. Determine the evolution factor of each particle based on the fitness value of each particle, the minimum fitness value of all current particles, and the maximum fitness value of all current particles;

[0046] S40. Determine the inertia weight of each particle according to the pre-determined initial inertia weight, end inertia weight, maximum number of iterations, and the evolution factor of each particle, and determine the learning factor of each particle according to the evolution factor of each particle and the maximum number of iterations;

[0047] S50. Update the velocity and position of each particle based on the inertia weight, learning factor, individual extreme value, and group extreme value of each particle;

[0048] S60. Randomly select a particle from each particle as the mutation particle, determine the particles whose fitness value of each particle is greater than the average fitness value of all current particles as the target particles, perform crossover and mutation operations on the mutation particle and each target particle according to a preset crossover and mutation function, calculate the fitness value of each target particle after crossover and mutation through the fitness function, and if the fitness value of each target particle after crossover and mutation is better than the fitness value before crossover and mutation, then replace the original target particle with the target particle after crossover and mutation;

[0049] S70. Determine whether the convergence condition is satisfied. If the convergence condition is not satisfied, return to step S20. If the convergence condition is satisfied, stop the search and output the current population extremum as the hyperparameters of the preset neural network.

[0050] Optionally, determining the evolution factor of each particle based on the fitness value of each particle, the minimum fitness value of all current particles, and the maximum fitness value of all current particles includes:

[0051] Determine the evolution factor of each particle through the following formula:

[0052]

[0053] where K(t, i) represents the evolution factor of particle i at the t-th iteration, f v (t, i) represents the fitness value of the i-th particle, f min (t) represents the minimum fitness value of all current particles, f max (t) represents the maximum fitness value of all current particles;

[0054] Determining the inertia weight of each particle based on the pre-determined initial inertia weight, end inertia weight, maximum number of iterations, and the evolution factor of each particle includes:

[0055] Determine the inertia weight of each particle through the following formula:

[0056]

[0057] where w(t, i) represents the inertia weight of particle i at the t-th iteration, w end represents the end inertia weight, whose value is 0.4, w start represents the initial inertia weight, whose value is 0.9, and T represents the maximum number of iterations;

[0058] Determining the learning factor of each particle based on the evolution factor of each particle and the maximum number of iterations includes:

[0059] Determine the learning factor of each particle through the following formula:

[0060]

[0061] Optionally, updating the velocity and position of each particle based on the inertia weight, learning factor, individual extremum, and population extremum of each particle includes:

[0062] Update the velocity of each particle through the following formula:

[0063] V i (t + 1) = w(t, i) * V i (t) + c1r1(Pbesti (t) - X i (t)) + c2r2(Gbest(t) - X i (t))

[0064] Update the positions of each particle through the following formula:

[0065] X i (t + 1) = X i (t) + V i (t + 1)

[0066] Among them, v i (t + 1) represents the velocity of particle i at time t + 1, v i (t) represents the velocity of particle i at time t, w is the inertia weight, c1 and c2 are learning factors, Pbest i (t) is the personal extreme value of particle i at the t-th iteration, Gbest(t) is the global extreme value of the entire particle swarm at the t-th iteration, r1 and r2 are random numbers, x i (t + 1) represents the position of particle i at time t + 1, x i (t) represents the position of particle i at time t;

[0067] The crossover and mutation function includes:

[0068]

[0069] Among them, fit(x) represents the fitness function, x(t, i) represents the target particle, x(t, k) represents the mutated particle, x(t, ii) represents the new individual generated after crossover and mutation, e is the Euler's constant, N(0, 1) follows a normal distribution, rand(0, 1) represents a random number between 0 and 1, and fav(t) represents the average fitness value of all current particles.

[0070] In the second aspect of the present application, a forage seed screening device suitable for a photovoltaic area is provided, including:

[0071] A data prediction module, configured to obtain the historical environmental parameters of the target photovoltaic area at each sampling moment within a specified historical period, and use the historical environmental parameters as input. The pre-trained environmental prediction model outputs the predicted environmental parameters of the target photovoltaic area at each sampling moment within a specified future period. The environmental prediction model is obtained by training a preset neural network with the historical environmental parameters of the target photovoltaic area and an improved optimization algorithm. The target photovoltaic area is the covered area of the solar panel;

[0072] A data matching module, configured to obtain the actual environmental parameters of each area to be monitored within the experimental area, determine the sampling moment that matches the current moment within the specified future period as the target moment, match the actual environmental parameters of each area to be monitored with the predicted environmental parameters at the target moment, and obtain the environmental parameter matching results of each area to be monitored. Among them, multiple forage seeds are pre-sown in each area to be monitored, and the forage seeds in each area to be monitored belong to the same seed category, and the forage seeds in all areas to be monitored belong to at least two seed categories;

[0073] An environmental adjustment module, configured to control an environmental adjustment device to adjust the environmental parameters of each area to be monitored based on the environmental parameter matching results of each area to be monitored;

[0074] A data analysis module, configured to obtain the forage images of each area to be monitored through an image acquisition device, determine the forage growth state data of each area to be monitored based on the forage images of each area to be monitored, and determine the seed category corresponding to the forage growth state data that meets the preset conditions among all the obtained forage growth state data as the target seed category.

[0075] The embodiments provided in this application have the following beneficial effects:

[0076] In this application, the neural network is trained through an improved optimization algorithm to obtain an environmental prediction model. Based on the environmental prediction model, the environmental parameters directly below the solar panel are simulated, and the sowing environment of different categories of seeds within the experimental area is adjusted through the simulated parameters. Furthermore, by monitoring and analyzing the growth conditions of different categories of seeds, the forage suitable for growing directly below the solar panel is screened out, effectively improving the prediction accuracy of the seed growth environment, ensuring the reliability of the simulation experiment, and improving the accuracy and efficiency of seed screening.

[0077] Other features and advantages of the embodiments or implementations of this application will be described in detail in the subsequent specific implementation part. BRIEF DESCRIPTION OF THE DRAWINGS

[0078] The drawings are used to provide a further understanding of the embodiments of this application, and constitute a part of the specification. They are used to explain the embodiments of this application together with the following specific implementation, but do not constitute a limitation to the embodiments of this application. In the drawings:

[0079] Figure 1 Schematically shows a flowchart of a method for screening forage seeds suitable for a photovoltaic area according to an implementation of this application;

[0080] Figure 2 Schematically shows a structural diagram of an image acquisition device according to an implementation of this application;

[0081] Figure 3 Schematically shows a schematic block diagram of a forage seed screening device suitable for a photovoltaic area according to an embodiment of the present application;

[0082] Figure 4 Schematically shows a schematic structural diagram of a terminal device according to an embodiment of the present application.

[0083] Description of reference numerals

[0084] 1 - First slide rail, 2 - Second slide rail, 3 - Third slide rail, 4 - Multispectral camera, 5 - Fixed bracket, 10 - Terminal device, 100 - Processor, 101 - Memory, 102 - Computer program. Specific embodiments

[0085] The following details the specific embodiments of the embodiments of the present application with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only for explaining and illustrating the embodiments of the present application, and are not used to limit the embodiments of the present application.

[0086] As Figure 1 shown, a forage seed screening method suitable for a photovoltaic area is provided in the first aspect of the present application, including:

[0087] S100. Obtain the historical environmental parameters of the target photovoltaic area at each sampling moment within a specified historical period. Using the historical environmental parameters as input, the pre-trained environmental prediction model outputs the predicted environmental parameters of the target photovoltaic area at each sampling moment within a specified future period. The environmental prediction model is obtained by training a preset neural network with the historical environmental parameters of the target photovoltaic area and an improved optimization algorithm. The target photovoltaic area is the covered area of the solar panel;

[0088] S200. Obtain the actual environmental parameters of each monitoring area within the experimental area, determine the sampling moment that matches the current moment within the specified future period as the target moment, and match the actual environmental parameters of each monitoring area with the predicted environmental parameters at the target moment to obtain the environmental parameter matching results of each monitoring area. Among them, multiple forage seeds are pre-sown in each monitoring area, and the forage seeds in each monitoring area belong to the same seed category. The forage seeds in all monitoring areas belong to at least two seed categories;

[0089] S300. Control the environmental adjustment device to adjust the environmental parameters of each monitoring area based on the environmental parameter matching results of each monitoring area;

[0090] S400. Obtain the forage images of each monitoring area through the image acquisition device, determine the forage growth state data of each monitoring area based on the forage images of each monitoring area, and determine that the seed category corresponding to the forage growth state data that meets the preset conditions among all the obtained forage growth state data is the target seed category.

[0091] Thus, in this application, the neural network is trained by an improved optimization algorithm to obtain an environmental prediction model. Based on the environmental prediction model, the environmental parameters directly below the solar panel are simulated, and the sowing environments of different types of seeds in the experimental area are adjusted through the simulated parameters. Furthermore, by monitoring and analyzing the growth conditions of different types of seeds, the forage grasses suitable for growing directly below the solar panel are screened out, effectively improving the prediction accuracy of the seed growth environment, ensuring the reliability of the simulation experiment, and improving the accuracy and efficiency of seed screening.

[0092] In step S100, the environmental parameters include light intensity, environmental temperature, and environmental humidity. Among them, the target photovoltaic area refers to the area covered by the solar panel in the photovoltaic array area to be sown. In this application, the experimental area is a pre-constructed laboratory. In the laboratory, each area to be monitored is a pre-divided planting area. For example, 1 area to be monitored can be 1 flowerpot. To simulate the environment of the area directly below the solar panel, this application can install air temperature and humidity sensors, total radiation sensors, soil moisture and conductivity sensors, soil pH sensors, etc. directly below the solar panel in the target photovoltaic area to collect environmental parameters such as air temperature and humidity, total radiation, soil moisture, conductivity, and soil pH value during the vegetation growth season, that is, from April to November, as the simulation parameters of the laboratory. To simulate the wild growth environment in the laboratory, the average values of each measured parameter during the day and at night from April to November can be calculated by statistically collecting the environmental parameters, so as to obtain the average values and then simulate the vegetation growth conditions. However, it is difficult to reflect the changes in environmental parameters at different time periods by simulating the vegetation growth conditions through the average values. For example, the environmental parameters at different time periods may change greatly due to meteorological conditions, and the changes in environmental parameters may have a greater impact on the growth of forage grasses. To more accurately simulate the environmental parameters of the target photovoltaic area and more accurately screen out the forage grass seeds suitable for growing below the solar panel, this application more accurately simulates the environmental parameters during the forage grass growth season through a neural network. In this application, the preset neural network can be any one of a BP neural network, a convolutional neural network, or a recurrent neural network. The light intensity, temperature, humidity, soil pH value, and soil conductivity of the target photovoltaic area in a certain historical period, such as in the past 1 or more years, are collected in advance. To more accurately simulate the environmental parameters, meteorological parameters during this historical period, such as wind speed, wind direction, etc., can also be added as inputs, and then the environmental parameters in the future period can be predicted more accurately. This application uses the collected historical data as input and outputs the predicted light intensity, temperature, humidity, soil pH value, and soil conductivity in the future period through a pre-trained prediction model. For example, the future period can be the corresponding period during the vegetation growth season, such as from April to November.

[0093] In step S200, the actual environmental parameters of each area to be monitored in the experimental area are obtained. Among them, the actual light intensity, actual environmental temperature, and actual environmental humidity parameters of each area to be monitored can directly adopt the light intensity, environmental temperature, and environmental humidity of the experimental area. The soil pH value and soil conductivity of each area to be monitored are separately monitored for each area to be monitored.

[0094] It can be understood that the sampling time of historical data is the same as that of actual data. For example, both are sampled once every 2 hours, or once every N hours. The smaller the interval time, the more real the simulated environmental parameters, but the computational workload will also increase. In order to reduce the computational workload while meeting the simulation requirements, the sampling time of this application can be set to once every 6 hours or 8 hours, so as to cover different typical periods of a day, such as morning, afternoon, evening, etc. Select common forage grass seeds in the northwest region in advance and conduct sowing experiments of forage grass seeds in different areas to be monitored. For example, conduct pot experiments in a simulation room. Select 10 common forage grass seeds in the northwest region in advance and sow them in each flower pot. Among them, each kind of forage grass is repeated in 5 flower pots, and the sowing amount in each flower pot is 50 seeds. Then place the flower pots in the simulation laboratory. After obtaining the predicted environmental parameters output by the prediction model, match each sampling time within the specified future period with the current time. If a match can be made, determine the current sampling time as the target time, obtain the actual environmental parameters of each area to be monitored at the current time, and match them with the predicted environmental parameters at the corresponding sampling time. For example, the sampling times in the future period are 0:00, 8:00, 16:00, 24:00 on April 1st,..., 0:00, 8:00, 16:00, 24:00 on November 30th. Then map the first day of the start of the simulation to April 1st, monitor the current time. When the current time is 0:00, collect the actual environmental parameters of each area to be monitored and match them with the predicted environmental parameters at 0:00 on April 1st to obtain the corresponding matching result.

[0095] It can be understood that in the present application, the environmental regulation device includes: a lighting device for adjusting the light intensity of the experimental area, a temperature regulation device for adjusting the environmental temperature of the experimental area, and a humidity regulation device for adjusting the environmental humidity of the experimental area. Among them, the lighting device can adopt existing LED lights or fluorescent lights, and simulate the change of light intensity at different time periods such as day and night by controlling the light intensity. The temperature regulation device can adopt an existing temperature regulator, and the humidity regulation device can adopt an existing humidification system, which is not limited herein. Each area to be monitored is provided with a corresponding PH adjustment device and a soil conductivity adjustment device for the soil PH value. Among them, the soil conductivity adjustment device and the PH adjustment device can be adjusted by means of drip irrigation of nutrient salt buffer solution and PH buffer solution. Among them, the soil conductivity and PH adjustment device are prior arts, which are not limited herein. In the present application, each flower pot is also equipped with a soil moisture sensor and a drip irrigation pipe. Among them, the soil moisture sensor is used to monitor the change of the water content in the flower pot, and the soil moisture monitoring system can automatically adjust the water content through the drip irrigation pipe according to the monitored change of the soil water content.

[0096] Then, in step S300, controlling the environmental regulation device to adjust the environmental parameters of each area to be monitored based on the matching results of the environmental parameters of each area to be monitored includes:

[0097] S310. Taking the actual light intensity of the experimental area as the actual light intensity of each area to be monitored, if the difference between the actual light intensity of each area to be monitored and the predicted light intensity is greater than the light intensity difference threshold, control the lighting device to adjust the actual light intensity of the experimental area until the difference between the actual light intensity of the experimental area and the predicted light intensity is not greater than the light intensity difference threshold. It can be understood that if the actual light intensity is less than the predicted light intensity, control the lighting device to increase the light intensity; conversely, control the lighting device to decrease the light intensity.

[0098] S320. Taking the actual environmental temperature of the experimental area as the actual environmental temperature of each area to be monitored, if the difference between the actual environmental temperature of each area to be monitored and the predicted environmental temperature is greater than the environmental temperature difference threshold, control the temperature regulation device to adjust the actual environmental temperature of the experimental area until the difference between the actual environmental temperature of the experimental area and the predicted environmental temperature is not greater than the environmental temperature difference threshold. Similarly, if the actual temperature is less than the predicted temperature, control the temperature regulator to increase the environmental temperature; conversely, decrease the environmental temperature.

[0099] S330. Use the actual environmental humidity of the experimental area as the actual environmental humidity of each area to be monitored. If the difference between the actual environmental humidity and the predicted environmental humidity of each area to be monitored is greater than the environmental humidity difference threshold, control the humidity adjustment device to adjust the environmental humidity of the experimental area until the difference between the environmental humidity of the experimental area and the predicted environmental humidity is not greater than the environmental humidity difference threshold. Similarly, if the actual humidity is less than the predicted humidity, control the humidifier to increase the environmental humidity; otherwise, decrease the environmental humidity.

[0100] S340. If the difference between the actual soil pH value of each area to be monitored and the preset soil pH reference value is greater than the soil pH difference threshold, control the pH adjustment device to adjust the soil pH value of each area to be monitored until the difference between the actual soil pH value and the soil pH reference value of each area to be monitored is not greater than the soil pH difference threshold. Similarly, if the actual soil pH value is less than the predicted soil pH value, control the soil pH value adjustment device to increase the soil pH value; otherwise, decrease the soil pH value.

[0101] S350. If the difference between the actual soil conductivity of each area to be monitored and the preset soil conductivity reference value is greater than the soil conductivity difference threshold, control the soil conductivity adjustment device to adjust the soil conductivity of each area to be monitored until the difference between the actual soil conductivity and the soil conductivity reference value of each area to be monitored is not greater than the soil conductivity difference threshold. Similarly, if the actual soil conductivity is less than the predicted soil conductivity, control the soil conductivity adjustment device to increase the soil conductivity; otherwise, decrease the soil conductivity.

[0102] Such as Figure 2As shown in the figure, the image acquisition device of the present application includes: at least one fixed bracket 5, a moving component, and a multispectral camera 4; wherein, the moving component includes a first slide rail 1, a second slide rail 2, and a third slide rail 3. Each slide rail can adopt an existing electric guide rail, such as a lead screw nut structure, to realize the sliding connection of each slide rail. The present application includes two fixed brackets 5. The fixed bracket 5 adopts a U-shaped structure. One end of the fixed bracket 5 is fixed on the ground of the experimental area, and the other end of the fixed bracket 5 is respectively fixedly connected to the first slide rail 1 and the second slide rail 2. The first slide rail 1 and the second slide rail 2 are arranged in parallel, and the plane where the first slide rail 1 and the second slide rail 2 are located is parallel to the ground; the third slide rail 3 is slidably connected to the first slide rail 1 and the second slide rail 2, and the third slide rail 3 is perpendicular to the first slide rail 1 and the second slide rail 2. The multispectral camera 4 is slidably connected to the third slide rail 3. In order to observe and record the growth status of forage grass under laboratory conditions, first, the first slide rail 1 and the second slide rail 2 are erected above each flower pot, and a multispectral lens mounting rod, that is, the third slide rail 3, is erected on the guide rail. The multispectral lens mounting rod can move along the electric guide rail; at the same time, the multispectral lens can move along the multispectral lens mounting rod. All movements can use a stepper motor as the drive, so that all flower pots can be image-captured by one image acquisition device. By presetting an automatic operation program for the stepper motor in advance, the multispectral lens is made to scan and take pictures regularly above each flower pot every day, obtaining multispectral images of the forage grass growth in the flower pots, and uploading the collected multispectral images to the server for image analysis through wireless transmission to obtain parameters such as the chlorophyll content, leaf area index, and vegetation index of the forage grass in each flower pot, and sorting each index. Among them, obtaining parameters such as the chlorophyll content, leaf area index, and vegetation index of plants by image analysis of multispectral images is a prior art, and no limitation is made here.

[0103] In step S400, after determining the forage grass growth status data of each area to be monitored, the method further includes: sorting the chlorophyll content, leaf area index, and vegetation index of the forage grass corresponding to all areas to be monitored from high to low respectively, to obtain a chlorophyll content queue, a leaf area index queue, and a vegetation index queue of the forage grass.

[0104] Then, the preset conditions include: the chlorophyll content of the forage grass in the currently monitored area is in the top n% of the chlorophyll content queue of the forage grass, the leaf area index of the currently monitored area is in the top n% of the leaf area index queue, and the vegetation index of the currently monitored area is in the top n% of the vegetation index queue. For example, 5 kinds of forage grass with chlorophyll content, leaf area index, and vegetation index all in the top 5 of the corresponding queues are selected as the target forage grass.

[0105] Alternatively, after determining the forage growth state data of each area to be monitored in step S400, the method further includes: sorting the chlorophyll content, leaf area index, and vegetation index of the forage corresponding to all areas to be monitored from high to low to obtain a chlorophyll content queue, a leaf area index queue, and a vegetation index queue of the forage; determining a first weight for the chlorophyll content of the forage, a second weight for the leaf area index, and a third weight for the vegetation index; determining a first score for the chlorophyll content of the forage in the current area to be monitored according to the preset forage growth state data scoring table and the position of the chlorophyll content of the forage in the current area to be monitored in the chlorophyll content queue, determining a second score for the leaf area index of the forage in the current area to be monitored according to the forage growth state data scoring table and the position of the leaf area index of the forage in the current area to be monitored in the leaf area index queue, and determining a third score for the vegetation index of the forage in the current area to be monitored according to the forage growth state data scoring table and the position of the vegetation index of the forage in the current area to be monitored in the vegetation index queue; performing weighted summation on the first score, the second score, and the third score based on the first weight, the second weight, and the third weight to obtain the forage growth state data scores of each area to be monitored, sorting the forage growth state data scores of each area to be monitored from high to low to obtain a forage growth state data score queue; the forage growth state data scoring table at least includes the first score corresponding to different positions of the chlorophyll content of the forage in the chlorophyll content queue, the second score corresponding to different positions of the leaf area index in the leaf area index queue, and the third score corresponding to different positions of the vegetation index in the vegetation index queue. Then, the preset condition includes: the forage growth state data score of the current area to be monitored is in the top n% of the forage growth state data score queue. The weights of the chlorophyll content, the leaf area index, and the vegetation index and the scores corresponding to different positions of different parameters in the corresponding queues are determined in advance, and the total score of the corresponding type of forage is determined by weighted summation, and the forage types with the top 5 total scores are used as the target forage.

[0106] In this application, the hyperparameters of the neural network model are optimized by an improved particle swarm algorithm. The existing particle swarm algorithm has problems such as insufficient search and being easily trapped in local optimal solutions. In order to balance the global search ability and the local search ability of the particle swarm algorithm, the improved optimization algorithm of this application includes:

[0107] S10. Initialize the particle swarm, determine the number of particles in the particle swarm and the initial positions of each particle, and map the hyperparameters of the preset neural network to different dimensions and different positions of each particle. For example, taking the BP neural network as an example, the hyperparameters to be optimized are the bias term and the weight, then the dimension of the particle is 2, and so on.

[0108] S20. Determine the fitness values of each particle based on a preset fitness function, and determine the minimum fitness value, the maximum fitness value, the average fitness value of all current particles, the individual extreme value of each particle, and the global extreme value of the population. Among them, the fitness function is constructed based on the output error of a preset neural network. For example, the fitness function can be constructed such that the smaller the output error of the neural network, the larger the fitness value. For example, the fitness function is constructed with the reciprocal of the output error.

[0109] S30. Determine the evolution factor of each particle based on the fitness value of each particle, the minimum fitness value of all current particles, and the maximum fitness value of all current particles.

[0110] S40. Determine the inertia weight of each particle based on a pre-determined initial inertia weight, an end inertia weight, a maximum number of iterations, and the evolution factor of each particle, and determine the learning factor of each particle based on the evolution factor of each particle and the maximum number of iterations.

[0111] Among them, the evolution factor of each particle is determined by the following formula:

[0112]

[0113] Among them, K(t, i) represents the evolution factor of particle i at the t-th iteration, f v (t, i) represents the fitness value of the i-th particle, f min (t) represents the minimum fitness value of all current particles, f max (t) represents the maximum fitness value of all current particles. It can be understood that the value range of K(t, i) is (0, 1). When its value is 0, it means that the current particle i has the strongest evolution degree at the t-th iteration, and when its value is 1, it means that the current particle i has the weakest evolution degree at the t-th iteration. By constructing the evolution factor, different inertia weight improvement strategies and learning factor improvement strategies can be formulated according to the evolution factor of the particle, realizing the adaptive adjustment of the inertia weight and the learning factor.

[0114] S50. Update the velocity and position of each particle based on the inertia weight, learning factor, individual extreme value, and global extreme value of each particle.

[0115] This application determines the inertia weight of each particle by the following formula:

[0116]

[0117] Among them, w(t, i) represents the inertia weight of particle i at the t-th iteration, w end represents the end inertia weight, and its value is 0.4, w start represents the initial inertia weight, and its value is 0.9, T represents the maximum number of iterations;

[0118] This application determines the learning factor of each particle through the following formula:

[0119]

[0120] This application updates the velocity of each particle through the following formula:

[0121] V i (t + 1) = w(t, i) * V i (t) + c1r1(Pbest i (t) - X i (t)) + c2r2(Gbest(t) - X i (t))

[0122] Update the position of each particle through the following formula:

[0123] X i (t + 1) = X i (t) + V i (t + 1)

[0124] Among them, v i (t + 1) represents the velocity of particle i at time t + 1, v i (t) represents the velocity of particle i at time t, w is the inertia weight, c1 and c2 are learning factors, Pbest i (t) is the individual extreme value of particle i at the t-th iteration, Gbest(t) is the global extreme value of the entire particle swarm at the t-th iteration, r1 and r2 are random numbers, x i (t + 1) represents the position of particle i at time t + 1, x i (t) represents the position of particle i at time t.

[0125] S60. Randomly select a particle from each particle as the mutant particle, determine the particles whose fitness value of each particle is greater than the average fitness value of all current particles as the target particles, perform crossover and mutation operations on the mutant particle and each target particle according to the preset crossover and mutation function, calculate the fitness value of each target particle after crossover and mutation through the fitness function, and if the fitness value of each target particle after crossover and mutation is better than the fitness value before crossover and mutation, then replace the original target particle with the target particle after crossover and mutation.

[0126] In this application, the crossover and mutation function includes:

[0127]

[0128] Among them, fit(x) represents the fitness function, x(t, i) represents the target particle, x(t, k) represents the mutated particle, x(t, ii) represents the new individual generated after crossover mutation, e is the Euler's constant, N(0, 1) follows a normal distribution, rand(0, 1) represents a random number between 0 and 1, and fav(t) represents the average fitness value of all current particles.

[0129] S70. Determine whether the convergence condition is satisfied. If the convergence condition is not satisfied, return to step S20. If the convergence condition is satisfied, stop the search and output the current population extreme value as the hyperparameter of the preset neural network. Among them, the convergence condition is to reach the maximum number of iterations or the output error of the neural network model is lower than the threshold.

[0130] As Figure 3 shown, in the second aspect of the present application, a forage seed screening device suitable for a photovoltaic area is provided, including:

[0131] A data prediction module, configured to obtain the historical environmental parameters of the target photovoltaic area at each sampling moment within a specified historical period, use the historical environmental parameters as input, and the pre-trained environmental prediction model outputs the predicted environmental parameters of the target photovoltaic area at each sampling moment within a specified future period. The environmental prediction model is obtained by training a preset neural network with the historical environmental parameters of the target photovoltaic area and an improved optimization algorithm. The target photovoltaic area is the covered area of the solar panel;

[0132] A data matching module, configured to obtain the actual environmental parameters of each area to be monitored within the experimental area, determine the sampling moment that matches the current moment within the specified future period as the target moment, and match the actual environmental parameters of each area to be monitored with the predicted environmental parameters at the target moment to obtain the environmental parameter matching results of each area to be monitored. Among them, multiple forage seeds are pre-sown in each area to be monitored, and the forage seeds within each area to be monitored belong to the same seed category. The forage seeds within all areas to be monitored belong to at least two seed categories;

[0133] An environmental adjustment module, configured to control the environmental adjustment device to adjust the environmental parameters of each area to be monitored based on the environmental parameter matching results of each area to be monitored;

[0134] A data analysis module, configured to obtain the forage images of each area to be monitored through an image acquisition device, determine the forage growth state data of each area to be monitored based on the forage images of each area to be monitored, and determine that the seed category corresponding to the forage growth state data that meets the preset conditions among all the obtained forage growth state data is the target seed category.

[0135] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the above division of each functional unit and module is used as an example. In actual applications, the above functions can be allocated to different functional units and modules according to needs, that is, the internal structure of the device is divided into different functional units or modules to complete all or part of the functions described above. Each functional unit and module in the embodiments can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit. The above integrated unit can be implemented in the form of hardware or in the form of a software functional unit. In addition, the specific names of each functional unit and module are only for the convenience of mutual distinction and do not limit the protection scope of this application. The specific working process of the units and modules in the above system can refer to the corresponding process in the foregoing method embodiments and will not be elaborated here.

[0136] In the third aspect of this application, a computer-readable storage medium is provided, storing a computer program that, when executed by a processor, causes the processor to execute the method for screening forage grass seeds suitable for a photovoltaic area as described above.

[0137] In the fourth aspect of this application, a terminal device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the method for screening forage grass seeds suitable for a photovoltaic area as described above is implemented.

[0138] As Figure 4 shown is a schematic diagram of the terminal device provided by the embodiment of this application. As Figure 4 shown, the terminal device 10 of this embodiment includes: a processor 100, a memory 101, and a computer program 102 stored in the memory 101 and executable on the processor 100. When the processor 100 executes the computer program 102, the steps in the foregoing method embodiments are implemented. Alternatively, when the processor 100 executes the computer program 102, the functions of each module / unit in the foregoing device embodiments are implemented.

[0139] Exemplarily, the computer program 102 can be divided into one or more modules / units. One or more modules / units are stored in the memory 101 and executed by the processor 100 to complete this application. One or more modules / units can be a series of computer program instruction segments capable of completing specific functions, and these instruction segments are used to describe the execution process of the computer program 102 in the terminal device 10.

[0140] The terminal device 10 can be a computing device such as a desktop computer, a notebook, a palm computer, and a cloud server. The terminal device 10 may include, but is not limited to, a processor 100 and a memory 101. Those skilled in the art can understand thatFigure 4 This is only an example of the terminal device 10, which does not constitute a limitation on the terminal device 10. It may include more or fewer components than those shown in the figure, or combine certain components, or different components. For example, the terminal device may also include input / output devices, network access devices, buses, etc.

[0141] The processor 100 may be a central processing unit (CPU), or may also be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0142] The memory 101 may be an internal storage unit of the terminal device 10, such as the hard disk or memory of the terminal device 10. The memory 101 may also be an external storage device of the terminal device 10, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the terminal device 10. Further, the memory 101 may also include both the internal storage unit and the external storage device of the terminal device 10. The memory 101 is used to store computer programs and other programs and data required by the terminal device 10. The memory 101 may also be used to temporarily store data that has been output or is to be output.

[0143] Those skilled in the art should understand that the embodiments of the present application may be provided as a method, a system, or a computer program product. Therefore, the present application may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memories, CD-ROMs, optical memories, etc.) containing computer-usable program codes.

[0144] It should also be noted that the term "comprise", "include" or any other variant thereof is intended to cover non-exclusive inclusion, such that a process, method, commodity or device comprising a series of elements not only includes those elements but also includes other elements not expressly listed, or further includes elements inherent to such process, method, commodity or device. Without further limitation, an element defined by the statement "comprising an..." does not exclude the presence of additional identical elements in the process, method, commodity or device comprising the element.

[0145] The above are only embodiments of the present application and are not used to limit the present application. For those skilled in the art, various changes and modifications can be made to the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included within the scope of the claims of the present application.

Claims

1. A method for screening forage grass seeds suitable for photovoltaic areas, characterized in that: include: Obtaining historical environmental parameters of a target photovoltaic area at each sampling moment in a specified historical period, taking the historical environmental parameters as input, and outputting predicted environmental parameters of the target photovoltaic area at each sampling moment in a specified future period through a pre-trained environmental prediction model, wherein the environmental prediction model is obtained by training a preset neural network with the historical environmental parameters of the target photovoltaic area and an improved optimization algorithm, and the target photovoltaic area is a covered area of ​​a solar panel; Acquire the actual environmental parameters of each area to be monitored in the experimental area, determine the sampling time matching the current time in the specified future period as the target time, match the actual environmental parameters of each area to be monitored with the predicted environmental parameters at the target time, and obtain the environmental parameter matching results of each area to be monitored, wherein each area to be monitored is pre-sown with multiple forage seeds, and the forage seeds in each area to be monitored belong to the same seed category, and the forage seeds in all areas to be monitored belong to at least two seed categories; Based on the matching results of the environmental parameters of each area to be monitored, the environmental adjustment device is controlled to adjust the environmental parameters of each area to be monitored; The grass images of each monitored area are obtained by an image acquisition device, and the grass growth status data of each monitored area is determined based on the grass images of each monitored area. The seed category corresponding to the grass growth status data that meets the preset conditions among all the obtained grass growth status data is determined as the target seed category.

2. The method for selecting forage grass seeds suitable for photovoltaic areas according to claim 1, characterized in that: The environmental parameters include: Light intensity, ambient temperature and ambient humidity; The preset neural network is a BP neural network, a convolutional neural network or a recurrent neural network; The environmental adjustment device comprises: An illumination device for adjusting the illumination intensity of the experimental area, a temperature adjustment device for adjusting the ambient temperature of the experimental area, and a humidity adjustment device for adjusting the ambient humidity of the experimental area; The image acquisition device comprises: at least one fixed support, a mobile assembly and a multispectral camera; The moving assembly includes a first slide rail, a second slide rail and a third slide rail; One end of the fixed bracket is fixed to the ground of the experimental area, and the other end of the fixed bracket is fixedly connected to the first slide rail and the second slide rail respectively, the first slide rail and the second slide rail are arranged in parallel, and the plane where the first slide rail and the second slide rail are located is parallel to the ground; The third slide rail is slidably connected to the first slide rail and the second slide rail, and the third slide rail is perpendicular to the first slide rail and the second slide rail, and the multi-spectral camera is slidably connected to the third slide rail.

3. The method for selecting forage grass seeds suitable for photovoltaic areas according to claim 2, characterized in that: Based on the matching results of the environmental parameters of each area to be monitored, the environmental adjustment device is controlled to adjust the environmental parameters of each area to be monitored, including: Taking the actual light intensity of the experimental area as the actual light intensity of each area to be monitored, if the difference between the actual light intensity of each area to be monitored and the predicted light intensity is greater than the light intensity difference threshold, controlling the lighting device to adjust the actual light intensity of the experimental area until the difference between the actual light intensity of the experimental area and the predicted light intensity is no greater than the light intensity difference threshold; Taking the actual ambient temperature of the experimental area as the actual ambient temperature of each area to be monitored, if the difference between the actual ambient temperature of each area to be monitored and the predicted ambient temperature is greater than the ambient temperature difference threshold, controlling the temperature adjustment device to adjust the actual ambient temperature of the experimental area until the difference between the actual ambient temperature of the experimental area and the predicted ambient temperature is no greater than the ambient temperature difference threshold; The actual ambient humidity of the experimental area is used as the actual ambient humidity of each monitored area. If the difference between the actual ambient humidity of each monitored area and the predicted ambient humidity is greater than the ambient humidity difference threshold, the humidity adjustment device is controlled to adjust the ambient humidity of the experimental area until the difference between the ambient humidity of the experimental area and the predicted ambient humidity is no greater than the ambient humidity difference threshold.

4. The method for selecting forage grass seeds suitable for photovoltaic areas according to claim 2, characterized in that: The environmental parameters also include soil pH value and soil conductivity; the environmental adjustment device also includes: A pH adjusting device for adjusting the pH value of the soil in the corresponding area to be monitored and a soil conductivity adjusting device for adjusting the soil conductivity in the corresponding area to be monitored, wherein the pH adjusting device and the soil conductivity adjusting device correspond to each area to be monitored one by one; Based on the matching results of the environmental parameters of each area to be monitored, the environmental adjustment device is controlled to adjust the environmental parameters of each area to be monitored, and further includes: If the difference between the actual soil pH value of each monitored area and the preset soil pH reference value is greater than the soil pH difference threshold, control the pH adjustment device to adjust the soil pH value of each monitored area until the difference between the actual soil pH value of each monitored area and the soil pH reference value is no greater than the soil pH difference threshold; If the difference between the actual soil conductivity of each area to be monitored and the preset soil conductivity reference value is greater than the soil conductivity difference threshold, the soil conductivity adjustment device is controlled to adjust the soil conductivity of each area to be monitored until the difference between the actual soil conductivity of each area to be monitored and the soil conductivity reference value is no greater than the soil conductivity difference threshold.

5. The method for selecting forage grass seeds suitable for photovoltaic areas according to claim 2, characterized in that: The forage growth status data includes: Chlorophyll content, leaf area index and vegetation index of forage grasses; After determining the grass growth status data of each area to be monitored, the method further includes: The chlorophyll content, leaf area index and vegetation index of the grasses corresponding to all the monitored areas are sorted from high to low to obtain the chlorophyll content queue, leaf area index queue and vegetation index queue of the grasses; The preset conditions include: The chlorophyll content of the grass in the current monitored area is in the top n% of the chlorophyll content queue of the grass, the leaf area index of the current monitored area is in the top n% of the leaf area index queue, and the vegetation index of the current monitored area is in the top n% of the vegetation index queue.

6. The method for selecting forage grass seeds suitable for photovoltaic areas according to claim 2, characterized in that: After determining the grass growth status data of each area to be monitored, the method further includes: The chlorophyll content, leaf area index and vegetation index of the grasses corresponding to all the monitored areas are sorted from high to low to obtain the chlorophyll content queue, leaf area index queue and vegetation index queue of the grasses; Determining a first weight of the chlorophyll content of the forage, a second weight of the leaf area index, and a third weight of the vegetation index; Determine a first score of the chlorophyll content of the grass in the current monitored area according to a preset grass growth status data scoring table and the position of the chlorophyll content of the grass in the current monitored area in the chlorophyll content queue of the grass, determine a second score of the leaf area index of the current monitored area according to the grass growth status data scoring table and the position of the leaf area index of the current monitored area in the leaf area index queue, and determine a third score of the vegetation index of the current monitored area according to the grass growth status data scoring table and the position of the vegetation index of the current monitored area in the vegetation index queue; Based on the first weight, the second weight and the third weight, a weighted sum is performed on the first score, the second score and the third score to obtain a grass growth status data score of each area to be monitored, and the grass growth status data scores of each area to be monitored are sorted from high to low to obtain a grass growth status data score queue; The forage growth status data scoring table at least includes a first score corresponding to different positions of the chlorophyll content of the forage in the chlorophyll content queue of the forage, a second score corresponding to different positions of the leaf area index in the leaf area index queue, and a third score corresponding to different positions of the vegetation index in the vegetation index queue; The preset conditions include: The forage growth status data score of the current area to be monitored is in the top n% of the forage growth status data score queue.

7. The method for selecting forage grass seeds suitable for photovoltaic areas according to claim 2, characterized in that: The improved optimization algorithm comprises: S10, initializing a particle swarm, determining the number of particles in the particle swarm and the initial position of each particle, and mapping the hyperparameters of the preset neural network to different dimensions and different positions of each particle; S20, determining the fitness value of each particle based on a preset fitness function, and determining the minimum fitness value of all current particles, the maximum fitness value of all current particles, the average fitness value of all current particles, the individual extreme value of each particle, and the group extreme value; the fitness function is constructed based on the output error of the preset neural network; S30, determining the evolution factor of each particle based on the fitness value of each particle, the minimum fitness value of all current particles, and the maximum fitness value of all current particles; S40, determining the inertia weight of each particle according to the predetermined initial inertia weight, the final inertia weight, the maximum number of iterations and the evolution factor of each particle, and determining the learning factor of each particle according to the evolution factor of each particle and the maximum number of iterations; S50, updating the speed and position of each particle based on the inertia weight, learning factor, individual extreme value and group extreme value of each particle; S60, randomly selecting a particle from each particle as a mutation particle, determining a particle whose fitness value of each particle is greater than the average fitness value of all current particles as a target particle, performing a crossover mutation operation on the mutation particle and each target particle according to a preset crossover mutation function, calculating the fitness value of each target particle after the crossover mutation by the fitness function, and if the fitness value of each target particle after the crossover mutation is better than the fitness value before the crossover mutation, replacing the original target particle with the target particle after the crossover mutation; S70, determining whether the convergence condition is met, if not, returning to step S20, if the convergence condition is met, stopping the search, and outputting the current population extreme value as the hyperparameter of the preset neural network.

8. The method for selecting forage grass seeds suitable for photovoltaic areas according to claim 7, characterized in that: The evolution factor of each particle is determined based on the fitness value of each particle, the minimum fitness value of all current particles, and the maximum fitness value of all current particles, including: The evolution factor of each particle is determined by the following formula: Among them, K(t, i) represents the evolution factor of particle i at the tth iteration, f v (t, i) represents the fitness value of the i-th particle, f min (t) represents the minimum fitness value of all particles at present, f max (t) represents the maximum fitness value of all particles at present; The inertia weight of each particle is determined based on the predetermined initial inertia weight, final inertia weight, maximum number of iterations and evolution factor of each particle, including: The inertia weight of each particle is determined by the following formula: Among them, w(t, i) represents the inertia weight of particle i at the tth iteration, w end Indicates the end inertia weight, its value is 0.4, w start represents the initial inertia weight, whose value is 0.9, and T represents the maximum number of iterations; The learning factor of each particle is determined based on the evolution factor of each particle and the maximum number of iterations, including: The learning factor of each particle is determined by the following formula:

9. The method for selecting forage grass seeds suitable for photovoltaic areas according to claim 8, characterized in that: Update the speed and position of each particle based on the inertia weight, learning factor, individual extreme value and group extreme value of each particle, including: Update the velocity of each particle using the following formula: V i (t+1)=w(t,i)*V i (t)+c1r1(Pbest i (t)-X i (t))+c2r2(Gbest(t)-X i (t)) Update the position of each particle using the following formula: X i (t+1)=X i (t)+V i (t+1) Among them, v i (t+1) represents the velocity of particle i at time t+1, v i (t) represents the velocity of particle i at time t, w is the inertia weight, c1 and c2 are learning factors, Pbest i (t) is the individual extreme value of particle i at the tth iteration, Gbest(t) is the group extreme value of the entire particle swarm at the tth iteration, r1 and r2 are random numbers, x i (t+1) represents the position of particle i at time t+1, x i (t) represents the position of particle i at time t; The cross-variogram function includes: Among them, fit(x) represents the fitness function, x(t, i) represents the target particle, x(t, k) represents the mutant particle, x(t, ii) represents the new individual generated after crossover mutation, e is the Euler constant, N(0,1) obeys the normal distribution, rand(0,1) represents a random number between 0 and 1, and fav(t) represents the average fitness value of all current particles.

10. A forage seed screening device suitable for photovoltaic areas, characterized in that: include: A data prediction module is configured to obtain historical environmental parameters of a target photovoltaic area at each sampling moment in a specified historical period, and to output predicted environmental parameters of the target photovoltaic area at each sampling moment in a specified future period using a pre-trained environmental prediction model, wherein the environmental prediction model is obtained by training a preset neural network using the historical environmental parameters of the target photovoltaic area and an improved optimization algorithm, and the target photovoltaic area is a covered area of ​​a solar panel; The data matching module is configured to obtain the actual environmental parameters of each area to be monitored in the experimental area, determine the sampling time matching the current time in the specified future period as the target time, match the actual environmental parameters of each area to be monitored with the predicted environmental parameters at the target time, and obtain the environmental parameter matching results of each area to be monitored, wherein each area to be monitored is pre-sown with a plurality of forage seeds, and the forage seeds in each area to be monitored belong to the same seed category, and the forage seeds in all areas to be monitored belong to at least two seed categories; An environment adjustment module is configured to control the environment adjustment device to adjust the environment parameters of each area to be monitored based on the matching results of the environment parameters of each area to be monitored; The data analysis module is configured to obtain grass images of each monitored area through an image acquisition device, determine the grass growth status data of each monitored area based on the grass images of each monitored area, and determine that the seed category corresponding to the grass growth status data that meets preset conditions among all the obtained grass growth status data is the target seed category.

Images