Automatic light supplement control method and system for vegetable cultivation

Abstract

The invention relates to the technical field of vegetable cultivation, and discloses an automatic light supplement control method and system for vegetable cultivation, and the method comprises the steps: determining the growth state of a vegetable crop based on crop image information and a crop growth model, and judging whether to execute a light supplement strategy or not according to a solar altitude, determining a backtracking light supplementing strategy or an initial light supplementing strategy according to the occurrence frequency of the growth state in the historical light supplementing database, determining target light supplementing data of the vegetable crops according to the number of the light supplementing data, and when the initial light supplementing strategy is determined, determining the target light supplementing data of the vegetable crops based on a recurrent neural network model, and analyzing the historical light supplement record to determine a light supplement swing factor of the light supplement lamp, adjusting the target light supplement data based on the light supplement swing factor, and performing light supplement on the vegetable crops according to the adjusted target light supplement data. According to the invention, the reliability of light supplement is ensured through the crop growth model and the light supplement swing factor.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of vegetable cultivation, in particular to an automatic light supplement control method and system for vegetable cultivation. BACKGROUND

[0002] Light is an important factor affecting the photosynthesis, growth cycle, yield and quality of vegetables. The growth of vegetables requires specific photosynthetic active radiation and light duration conditions. For example, leaf vegetables require 8-10 hours of moderate light per day to promote leaf differentiation. However, natural light is affected by factors such as season, region and weather. Therefore, light supplement lamps have become an important device for vegetable cultivation in facilities such as greenhouses.

[0003] In the face of vegetable cultivation, light supplement lamps usually rely on human experience to set parameters. The parameters set by humans depend on the experience of planting personnel. Due to individual cognitive differences, the parameters set by different planting personnel do not match the actual needs of vegetables. Relying on human experience cannot dynamically adjust the light supplement lamp according to the situation of vegetable cultivation, and lacks a feedback regulation mechanism, resulting in insufficient reliability of vegetable cultivation.

[0004] Therefore, it is necessary to design an automatic light supplement control method and system for vegetable cultivation to solve the problems existing in the current technology. SUMMARY

[0005] In view of this, the present application provides an automatic light supplement control method and system for vegetable cultivation, aiming to solve the problem that the parameters set by different planting personnel do not match the actual needs of vegetables, relying on human experience cannot dynamically adjust the light supplement lamp according to the situation of vegetable cultivation, and lacks a feedback regulation mechanism, resulting in insufficient reliability of vegetable cultivation.

[0006] In one aspect, the present application provides an automatic light supplement control method for vegetable cultivation, comprising:

[0007] Obtaining crop image information of a vegetable crop, determining the growth state of the vegetable crop based on the crop image information and a crop growth model, and determining whether to execute a light supplement strategy according to the solar elevation angle;

[0008] When it is determined to execute the light supplement strategy, searching in a historical light supplement database based on the growth state, and determining a backtracking light supplement strategy or an initial light supplement strategy according to the number of occurrences of the growth state in the historical light supplement database;

[0009] When the backtracking light supplement strategy is determined, corresponding light supplement data is determined in the historical light supplement database based on the growth state, and target light supplement data of the vegetable crop is determined according to the number of the light supplement data; when the initial light supplement strategy is determined, the target light supplement data of the vegetable crop is determined based on a recurrent neural network model.

[0010] The historical light supplement records of the light supplement lamp are collected, and the light supplement swing factor of the light supplement lamp is determined by analyzing the historical light supplement records; the target light supplement data is adjusted based on the light supplement swing factor, and the vegetable crop is light supplemented according to the adjusted target light supplement data.

[0011] Further, when the growth state of the vegetable crop is determined based on the crop image information and the crop growth model, the following steps are further included:

[0012] A growth data set of the vegetable crop is obtained, and the growth data set is divided into a training set and a test set; a model parameter is found based on grid search; and a convolutional neural network model is established based on the model parameter;

[0013] The convolutional neural network model is trained based on the training set; the test set is substituted into the trained convolutional neural network model to determine a prediction accuracy; and the crop growth model is determined based on the prediction accuracy.

[0014] Further, when the growth state of the vegetable crop is determined based on the crop image information and the crop growth model, the following steps are further included:

[0015] When the prediction accuracy is greater than or equal to a prediction accuracy threshold, the current trained convolutional neural network model is determined as the crop growth model;

[0016] When the prediction accuracy is less than the prediction accuracy threshold, the amplitude of the change of the current trained convolutional neural network model in the gradient direction is reduced, and the training is continued;

[0017] The crop image information is substituted into the crop growth model to determine the growth state of the vegetable crop.

[0018] Further, when it is determined whether to execute the light supplement strategy according to the solar elevation angle, the following steps are further included:

[0019] A solar elevation angle threshold is preset, and the solar elevation angle threshold and the solar elevation angle are compared;

[0020] When the solar elevation angle is greater than or equal to the solar elevation angle threshold, it is determined that the light supplement strategy is not executed;

[0021] When the solar elevation angle is less than the solar elevation angle threshold, it is determined that the light supplement strategy is executed.

[0022] The solar elevation angle threshold is determined based on the growth state.

[0023] Further, in determining the backtracking light supplement strategy or the initial light supplement strategy according to the number of occurrences of the growth state in the historical light supplement database, comprising:

[0024] The historical light supplement database comprises a plurality of historical growth states, a plurality of historical light supplement data, and the number of occurrences of each historical growth state, and each historical growth state corresponds to a historical light supplement data, the historical light supplement data comprising historical light supplement intensity and historical light supplement angle;

[0025] When the growth state has the same historical growth state in the historical light supplement database, and the corresponding number of occurrences is greater than or equal to the number of occurrences threshold, the backtracking light supplement strategy is determined;

[0026] When the growth state does not have the same historical growth state in the historical light supplement database, or when the growth state has the same historical growth state in the historical light supplement database, and the corresponding number of occurrences is less than the number of occurrences threshold, the initial light supplement strategy is determined.

[0027] Further, in determining the corresponding light supplement data of the vegetable crop based on the growth state in the historical light supplement database, and determining the target light supplement data of the vegetable crop according to the number of light supplement data, comprising:

[0028] Extracting the same historical growth state of the growth state in the historical light supplement database, and determining the mean value of the historical light supplement data corresponding to each historical growth state as the target light supplement data of the vegetable crop.

[0029] Further, in determining the target light supplement data of the vegetable crop based on the recurrent neural network model, comprising:

[0030] The recurrent neural network model is pre-trained, and the solar irradiation trajectory is determined based on a solar tracking algorithm;

[0031] The input layer of the recurrent neural network model is used to receive the solar irradiation trajectory and the growth state;

[0032] The hidden layer of the recurrent neural network model comprises a first LSTM layer, a second LSTM layer, a first full connection layer and a second full connection layer;

[0033] The number of neurons of the first LSTM layer is set to 128, the number of neurons of the second LSTM layer is set to 64, and the first full connection layer and the second full connection layer are used to compress features;

[0034] The output layer of the recurrent neural network model comprises a light supplement intensity output layer and a light supplement angle output layer;

[0035] The number of neurons of the light supplement intensity output layer and the light supplement angle output layer is 1;

[0036] The target light supplement data of the vegetable crop is determined based on the light supplement intensity output layer and the light supplement angle output layer.

[0037] Further, when collecting the historical light supplement record of the light supplement lamp and analyzing the historical light supplement record to determine the light supplement swing factor of the light supplement lamp, the method comprises:

[0038] The swing behavior and the non-swing behavior in the historical light supplement record are determined, and the number of swing behaviors and the number of non-swing behaviors are obtained;

[0039] The number ratio m of the number of swing behaviors and the number of non-swing behaviors is obtained;

[0040] A first preset light supplement swing factor, a second preset light supplement swing factor and a third preset light supplement swing factor are preset;

[0041] When m≤1, the first preset light supplement swing factor is determined as the light supplement swing factor of the light supplement lamp;

[0042] When 1

[0043] When 1.5

[0044] The first preset light supplement swing factor is smaller than the second preset light supplement swing factor, and the second preset light supplement swing factor is smaller than the third preset light supplement swing factor.

[0045] Further, when adjusting the target light supplement data based on the light supplement swing factor, the method comprises:

[0046] The target light supplement data is in a positive relationship with the light supplement swing factor.

[0047] Compared with the prior art, the present application has the beneficial effects that: the growth state of the vegetable crop is determined based on crop image information and a crop growth model, ensuring the matching degree of the light supplementing parameters and the growth demand of the vegetable crop, avoiding the disconnection between light supplementing and the actual demand of the vegetable crop caused by relying on human experience, realizing dynamic response to environmental changes by judging whether to execute the light supplementing strategy according to the solar elevation angle, determining the backtracking light supplementing strategy or the initial light supplementing strategy according to the number of occurrences of the growth state in the historical light supplementing database, and determining the target light supplementing data by reusing mature experience or with the help of a recurrent neural network model, which can not only guarantee the stability of light supplementing under normal growth conditions, but also adapt to special or new growth conditions, ensuring the flexibility and adaptability of light supplementing. The historical light supplementing records of the light supplementing lamp are collected to determine the light supplementing swing factor, forming a closed-loop feedback of growth state monitoring, light supplementing execution, historical record analysis and parameter adjustment, and dynamically controlling the light supplementing effect according to the growth state of the vegetable crop and the fluctuation amplitude of light supplementing, thereby improving the reliability of light supplementing and providing stable photosynthetically active radiation and light duration for the vegetable crop.

[0048] In another aspect, the present application also provides an automatic light supplementing control system for vegetable cultivation, which is used for applying the automatic light supplementing control method for vegetable cultivation, comprising:

[0049] The acquisition and judgment module is configured to acquire crop image information of the vegetable crop, determine the growth state of the vegetable crop based on the crop image information and a crop growth model, and judge whether to execute a light supplementing strategy according to the solar elevation angle;

[0050] The strategy determination module is configured to, when it is determined to execute the light supplementing strategy, search in a historical light supplementing database based on the growth state, and determine a backtracking light supplementing strategy or an initial light supplementing strategy according to the number of occurrences of the growth state in the historical light supplementing database;

[0051] The light supplementing determination module is configured to, when it is determined to be the backtracking light supplementing strategy, determine corresponding light supplementing data in the historical light supplementing database based on the growth state, and determine the target light supplementing data of the vegetable crop according to the number of the light supplementing data, and, when it is determined to be the initial light supplementing strategy, determine the target light supplementing data of the vegetable crop based on a recurrent neural network model;

[0052] The light supplementing control module is configured to collect historical light supplementing records of a light supplementing lamp, analyze the historical light supplementing records to determine a light supplementing swing factor of the light supplementing lamp, adjust the target light supplementing data based on the light supplementing swing factor, and supplement light to the vegetable crop according to the adjusted target light supplementing data.

[0053] It can be understood that the above-mentioned automatic light supplementing control method and system for vegetable cultivation have the same beneficial effects, which will not be described here again. Attached Figure Description

[0054] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings:

[0055] Figure 1 A flowchart illustrating an automated supplemental lighting control method for vegetable cultivation, provided in an embodiment of the present invention;

[0056] Figure 2 This is a functional block diagram of an automated supplemental lighting control system for vegetable cultivation, provided as an embodiment of the present invention. Detailed Implementation

[0057] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the disclosure to those skilled in the art. It should be noted that, unless otherwise specified, embodiments and features in the embodiments of the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0058] In some embodiments of this application, see Figure 1 As shown, an automated supplemental lighting control method for vegetable cultivation includes:

[0059] S100: Acquire crop image information of vegetable crops, determine the growth status of vegetable crops based on crop image information and crop growth model, and determine whether to implement supplemental lighting strategy based on solar altitude angle.

[0060] S200: When it is determined to execute the supplemental lighting strategy, the historical supplemental lighting database is searched based on the growth status, and the retrospective supplemental lighting strategy or the initial supplemental lighting strategy is determined according to the number of times the growth status appears in the historical supplemental lighting database.

[0061] S300: When the retrospective supplemental lighting strategy is determined, the corresponding supplemental lighting data is determined in the historical supplemental lighting database based on the growth status, and the target supplemental lighting data for vegetable crops is determined according to the number of supplemental lighting data. When the initial supplemental lighting strategy is determined, the target supplemental lighting data for vegetable crops is determined based on the recurrent neural network model.

[0062] S400: Collect the historical light supplement record of the light supplement lamp, analyze the historical light supplement record to determine a light supplement swing factor of the light supplement lamp, adjust the target light supplement data based on the light supplement swing factor, and supplement light for the vegetable crops according to the adjusted target light supplement data.

[0063] Specifically, the vegetable crops are cultivated through facilities such as greenhouses, the crop image information of the vegetable crops is collected according to image collection devices such as high-definition cameras, the crop image information reflects visual features such as the plant height, the number of leaves, and the stem thickness of the vegetable crops, the growth state of the vegetable crops is extracted from the crop image information by using a crop growth model, the growth state distinguishes different stages such as the seedling stage, the vegetative growth stage, and the reproductive growth stage of the vegetable crops, and contains data such as the leaf color, the degree of expansion, and the leaf thickness, and the growth state directly reflects the growth of the vegetable crops, providing a data basis for subsequent matching of appropriate parameters of the light supplement lamp. The solar elevation angle is a key factor for determining the intensity and effective duration of natural light received by the facilities such as greenhouses. When the solar elevation angle is larger, such as at noon, the path of sunlight through the atmosphere is shorter, the energy loss such as absorption and scattering of light by the atmosphere is less, and the photosynthetically active radiation reaching the surface of the vegetable crops, i.e., the intensity of light available for photosynthesis of the vegetable crops, is higher. Conversely, when the solar elevation angle is smaller, such as in the morning, in the evening, or in winter, the path of light through the atmosphere is longer, the energy loss increases, and the intensity of the photosynthetically active radiation actually reaching the crops is reduced, and may even be lower than the minimum threshold required by the vegetable crops at the current growth stage. The actual supply capacity of natural light is determined by monitoring the solar elevation angle, and whether it meets the demand of the vegetable crops for the photosynthetically active radiation and the light duration at the current growth state is analyzed, so as to determine whether to execute the light supplement strategy. When it is determined that the light supplement strategy needs to be executed, the historical light supplement database is searched first, and the backtracking light supplement strategy or the initial light supplement strategy is dynamically adopted according to the matching condition of the current growth state and the records in the historical light supplement database. When the backtracking light supplement strategy is determined, it is indicated that there are a certain number of historical records for the growth state in the historical light supplement database, so that the corresponding light supplement data is determined in the historical light supplement database, and the target light supplement data of the vegetable crops is determined according to the number of light supplement data. When the initial light supplement strategy is determined, it is indicated that there are few or no historical records for the growth state in the historical light supplement database, so that the data in the historical light supplement database is directly used, which is easy to cause adjustment deviation. Therefore, the recurrent neural network model is used to generate the target light supplement data in combination with the time sequence rule of the vegetable growth and the photosynthetic demand characteristics. Moreover, the light supplement lamp of the facilities such as greenhouses may swing due to the influence of weather such as sand and precipitation. The historical light supplement record of the light supplement lamp is collected, the swing condition in the historical light supplement record is analyzed, so as to determine a light supplement swing factor reflecting the light supplement stability, the target light supplement data is fine-tuned according to the light supplement swing factor, and the vegetable crops are finally supplemented with light according to the adjusted target light supplement data, ensuring the reliability of the light supplement.

[0064] It can be understood that individual cognitive differences will cause the set parameters of the light supplement lamp to be inconsistent with the actual needs of the vegetables, and determining the growth state of the vegetable crops through crop image information and a crop growth model eliminates human subjective bias, and the necessity of light supplement is determined in combination with the solar elevation angle, effectively avoiding energy waste when natural light is sufficient, and also preventing light supplement omission when natural light is insufficient to affect the growth of the vegetables. The historical light supplement database not only utilizes past effective data, but also covers the growth state with no historical data or a small amount of historical data through a recurrent neural network, further improving the reliability of light supplement. According to the light supplement swing factor, a feedback adjustment mechanism is established to avoid parameter deviation of the light supplement lamp caused by environmental changes, improve the accuracy of light supplement, and reduce the dependence on human experience, thereby ensuring the photosynthesis of the vegetables. Even if the natural light is affected by seasons, regions, and weather, stable light supplement can also ensure the growth of the vegetables.

[0065] In some embodiments of the present application, when the growth state of the vegetable crops is determined based on the crop image information and the crop growth model, the growth data set of the vegetable crops is obtained, and the growth data set is divided into a training set and a test set. The model parameters are found according to the grid search, the convolutional neural network model is established based on the model parameters, the convolutional neural network model is trained based on the training set, the test set is substituted into the trained convolutional neural network model to determine the prediction accuracy, and the crop growth model is determined based on the prediction accuracy.

[0066] In some embodiments of the present application, when the growth state of the vegetable crops is determined based on the crop image information and the crop growth model, the method further comprises: when the prediction accuracy is greater than or equal to the prediction accuracy threshold, the current trained convolutional neural network model is determined as the crop growth model; and when the prediction accuracy is less than the prediction accuracy threshold, the amplitude of the change of the current trained convolutional neural network model in the gradient direction is reduced, and the training is continued. The crop image information is substituted into the crop growth model to determine the growth state of the vegetable crops.

[0067] Specifically, a growth dataset of the vegetable crop is obtained, the growth dataset including multi-dimensional images of different growth stages and different health conditions of the vegetable crop, for example, whole cycle images from the seedling stage to the reproductive stage, and close-up images of different states of the leaves such as normal, yellowing, wilting, etc. The growth dataset is divided into a training set for model training and a test set for verifying performance, and the division ratio is usually 7:3. Model parameters suitable for image feature analysis are screened out through grid search, including learning rate, activation function, and regularization parameter, etc. A convolutional neural network model (CNN) is built based on these parameters. The convolutional neural network model is good at extracting detailed features in images, which meets the processing requirements of crop image information. Then the model is trained with the training set. After the training is completed, the test set is input into the model to calculate the prediction accuracy. The prediction accuracy is an important factor to evaluate the performance of the model. The prediction accuracy threshold is preferably 0.85. If the prediction accuracy reaches or exceeds the prediction accuracy threshold, it is considered that the performance of the model is relatively stable, and the training of the model can be stopped. The trained model is determined as the crop growth model. If the prediction accuracy threshold is not reached, the change amplitude of the model in the gradient direction is reduced (the learning rate is improved to slow down the parameter fluctuation in the training process), and the training is continued until the prediction accuracy meets the standard. Finally, the growth state of the vegetable crop is determined according to the crop growth model.

[0068] It can be understood that the grid search can find the optimal model parameters by screening, avoiding subjective bias of human experience. The application of the convolutional neural network model is targeted to extract crop image information, effectively capturing the growth state of the vegetable crop, thereby ensuring the accuracy of the light supplement control.

[0069] In some embodiments of the present application, when determining whether to perform the light supplement strategy according to the solar elevation angle, the following steps are included: a solar elevation angle threshold is set in advance, and the solar elevation angle threshold is compared with the solar elevation angle. When the solar elevation angle is greater than or equal to the solar elevation angle threshold, it is determined that the light supplement strategy is not performed. When the solar elevation angle is less than the solar elevation angle threshold, it is determined that the light supplement strategy is performed. The solar elevation angle threshold is determined based on the growth state.

[0070] Specifically, a corresponding solar elevation angle threshold is preset based on the current growth state of the vegetable crop. Due to the difference in the demand for photosynthetically active radiation and light duration of the vegetable in different growth states, the demand for light in the reproductive growth period is generally higher than that in the seedling stage. Moreover, the natural light is affected by the law of the sun. When the solar elevation angle is larger, such as at noon, the path of sunlight through the atmosphere is shorter, the energy loss of light absorption and scattering of the atmosphere is less, and the light intensity available for photosynthesis of the vegetable is higher. Conversely, when the solar elevation angle is smaller, such as in the morning, evening or winter, the path of light through the atmosphere is longer, the energy loss increases, and the photosynthetically active radiation intensity actually reaching the crop decreases. Therefore, the solar elevation angle threshold is matched with the demand of the vegetable in different growth stages, and the solar elevation angle threshold is bound with the growth state. The solar elevation angle threshold is not specifically limited herein. The solar elevation angle in the environment is monitored in real time, and is compared with the solar elevation angle threshold. If the current solar elevation angle is greater than or equal to the solar elevation angle threshold, it indicates that the intensity, effective radiation and duration of the natural light at this time can meet the demand of the current growth state of the vegetable, and it is determined that the light supplement strategy is not executed. If the current solar elevation angle is less than the solar elevation angle threshold, it indicates that the natural light supply is insufficient, and the vegetable cannot efficiently carry out photosynthesis, and it is determined that the light supplement strategy is executed.

[0071] It can be understood that, in the conventional manner, the light supplement is determined by subjective judgment without considering the solar elevation angle, or a fixed solar elevation angle threshold is used without considering the difference in the demand of the vegetable in different growth states, resulting in energy waste due to light supplement in the case of sufficient natural light, or missed light supplement due to insufficient light, which affects the growth of the vegetable. By binding the solar elevation angle threshold with the growth state, the light supplement timing can take into account both the natural light supply capacity and the actual demand of the crop, reduce the energy consumption of invalid light supplement, and prevent the growth from being hindered due to insufficient light, thereby further ensuring the effectiveness of the light supplement control.

[0072] In some embodiments of the present application, when the backtracking light supplement strategy or the initial light supplement strategy is determined according to the number of occurrences of the growth state in the historical light supplement database, the historical light supplement database includes a plurality of historical growth states, a plurality of historical light supplement data, and the number of occurrences of each historical growth state, and each historical growth state corresponds to a historical light supplement data. The historical light supplement data includes a historical light supplement intensity and a historical light supplement angle. When the same historical growth state exists in the historical light supplement database, and the corresponding number of occurrences is greater than or equal to the number of occurrences threshold, the backtracking light supplement strategy is determined. When the same historical growth state does not exist in the historical light supplement database, or when the same historical growth state exists in the historical light supplement database, and the corresponding number of occurrences is less than the number of occurrences threshold, the initial light supplement strategy is determined.

[0073] In some embodiments of the present application, when the corresponding light supplement data is determined in the historical light supplement database based on the growth state, and the target light supplement data of the vegetable crop is determined according to the number of light supplement data, it includes: extracting the historical growth states that are the same as the growth state in the historical light supplement database, and determining the mean value of the historical light supplement data corresponding to each historical growth state as the target light supplement data of the vegetable crop.

[0074] Specifically, the historical light supplement database stores a plurality of sets of historical information, including the historical growth state of the vegetable crop and the historical light supplement data corresponding to each historical growth state, and the number of occurrences of each historical growth state in the historical light supplement database. When determining the light supplement strategy, the growth state of the current vegetable crop is compared with the historical growth state in the historical light supplement database, the occurrence threshold is preferably 5 times, if there is a same historical growth state, and the number of occurrences of the historical growth state is greater than or equal to the occurrence threshold, it is determined to use the backtracking light supplement strategy, if the same historical growth state is not found, or the same historical growth state is found, and the number of occurrences is less than the occurrence threshold, it is determined to use the initial light supplement strategy. When the backtracking light supplement strategy is used, all historical growth states that are the same as the current growth state are extracted from the historical light supplement database, and the mean value of all historical light supplement data corresponding to these historical growth states is calculated, and finally the mean value is determined as the target light supplement data of the current vegetable crop. The target light supplement data is the light supplement intensity and the light supplement angle of the light supplement lamp.

[0075] It can be understood that the traditional method relies on human experience to set the light supplement intensity and the light supplement angle of the light supplement lamp, and does not use historical data. At the same time, different experienced planting personnel will cause deviation in the set parameters. Through the matching of the historical light supplement database and the growth state, the backtracking light supplement strategy can rely on objective data to directly use these data to determine the target light supplement data, thereby ensuring the reliability and consistency of the light supplement operation. Through the automatic adjustment driven by data, the dependence on human experience and intuition is reduced, and the uncertainty and light supplement deviation risk caused by human judgment are reduced, thereby improving the automation level and accuracy of light supplement control.

[0076] In some embodiments of the present application, when determining the target light supplement data of the vegetable crop based on the recurrent neural network model, the following steps are included: pre-training the recurrent neural network model, and determining the sun irradiation track based on a sun tracking algorithm; the input layer of the recurrent neural network model is used to receive the sun irradiation track and the growth state; the hidden layer of the recurrent neural network model includes a first LSTM layer, a second LSTM layer, a first full connection layer and a second full connection layer; the number of neurons of the first LSTM layer is set to 128, and the number of neurons of the second LSTM layer is set to 64; the first full connection layer and the second full connection layer are used to compress features; the output layer of the recurrent neural network model includes a light supplement intensity output layer and a light supplement angle output layer; the number of neurons of the light supplement intensity output layer and the light supplement angle output layer is 1; and the target light supplement data of the vegetable crop is determined based on the light supplement intensity output layer and the light supplement angle output layer.

[0077] Specifically, if the same historical growth state is not found, or the same historical growth state is found but the number of occurrences is less than the number of occurrence threshold, it is determined to use an initial light supplement strategy. The initial light supplement strategy determines the target light supplement data of the vegetable crop based on a recurrent neural network model. The recurrent neural network model is pre-trained, and the training process is consistent with that of the convolutional neural network model, which will not be repeated here. The recurrent neural network model has the ability to integrate time series information and analyze crop growth needs. The sun tracking algorithm is used to capture and determine the sun's irradiation track at different times. This track can intuitively reflect the dynamic change law of natural light. The process of capturing and determining the sun's irradiation track at different times based on the sun tracking algorithm is lengthy and mature, and will not be described in detail here. The sun irradiation track and the current growth state of the vegetable are input into the recurrent neural network model. After the input layer receives these two types of information, it is transmitted to the hidden layer. The two LSTM layers are responsible for processing the time series characteristics of the sun irradiation track, such as the trend of the sun's position change within a day, accurately capturing the fluctuation law of natural light over time. The two full connection layers then integrate and compress the features processed by the LSTM layers, remove redundant information to simplify calculations, and the output layer is divided into a light supplement intensity output layer and a light supplement angle output layer. The two output layers output the prediction results of the light supplement parameters, and the target light supplement data can be determined based on the results of the two output layers.

[0078] It can be understood that the sun irradiation track has certain time series characteristics, the recurrent neural network model is good at processing time series information and effectively associating the corresponding relationship between light changes at different times and crop light supplement needs. By inputting the sun irradiation track and the growth state, the model can balance the supply of natural light and the actual needs of the crop, avoid light supplement deviation caused by single factor judgment, reduce the waste of light supplement energy, and at the same time, strengthen the reliability and adaptability of automatic light supplement control.

[0079] In some embodiments of the present application, when the light supplement lamp history light supplement record is collected and analyzed to determine the light supplement swing factor of the light supplement lamp, it includes: determining the swing behavior and non-swing behavior in the history light supplement record, and obtaining the swing number of the swing behavior and the non-swing number of the non-swing behavior, obtaining the number ratio m of the swing number and the non-swing number, and pre-setting the first preset light supplement swing factor, the second preset light supplement swing factor and the third preset light supplement swing factor. When m≤1, the first preset light supplement swing factor is determined as the light supplement swing factor of the light supplement lamp. When 1

[0080] Specifically, the first preset light supplement swing factor is preferably 1.1, the second preset light supplement swing factor is preferably 1.3, and the third preset light supplement swing factor is preferably 1.5. The past light supplement records of the light supplement lamp are collected, and the swing behavior and non-swing behavior are distinguished from the records, wherein the swing behavior refers to the unstable fluctuation of the light supplement angle in the light supplement process, for example, the wind speed disturbance of the light supplement angle of the light supplement lamp caused by weather such as sand and rain. The non-swing behavior refers to the case that the light supplement angle remains stable and has no obvious fluctuation in the light supplement process. The swing number of the swing behavior and the non-swing number of the non-swing behavior are counted, and the number ratio m is calculated. The larger the number ratio is, the higher the swing degree of the light supplement lamp affected by the environment is. When m≤1, the swing degree of the light supplement lamp affected by the environment is the smallest, and the smallest first preset light supplement swing factor is determined as the light supplement swing factor of the light supplement lamp. When 1

[0081] It can be understood that when the light supplement lamp is actually working, it is affected by environmental air flow and other factors, and the light supplement parameters will deviate from the ideal light supplement condition. By distinguishing the swing behavior and non-swing behavior and calculating the number ratio m, the unstable degree of the target light supplement data is effectively quantified, and different preset light supplement swing factors are used to compensate for the corresponding fluctuation level, which ensures the actual light supplement state of the light supplement lamp, compensates for the defects of the traditional light supplement lacking fluctuation correction, reduces the influence of the actual light supplement deviation on the growth of vegetables, and further improves the accuracy and reliability of the light supplement control, thereby ensuring that the vegetables obtain stable photosynthetically active radiation.

[0082] In some embodiments of the present application, when adjusting the target light compensation data based on the light compensation swing factor, the target light compensation data is in a proportional relationship with the light compensation swing factor.

[0083] Specifically, the target light compensation data is adjusted according to the light compensation swing factor. Assuming that the light compensation intensity of the target light compensation data is L, the light compensation angle is H, and the light compensation swing factor is P, the adjusted target light compensation data is determined as L*P and H*P. When a wider light compensation angle and stronger light compensation intensity are needed to compensate for the fluctuation of the light compensation lamp, the adjusted target light compensation data will be increased synchronously with the increase of the light compensation swing factor. By establishing a proportional relationship between the target light compensation data and the light compensation swing factor, the light compensation intensity and the light compensation angle of the light compensation lamp are accurately controlled, and the reliability of light compensation is ensured.

[0084] In summary, the beneficial effects of the present application are that the growth state of the vegetable crop is determined based on the crop image information and the crop growth model, ensuring the matching degree of the light compensation parameters and the growth demand of the vegetable crop, avoiding the disconnection between the light compensation and the actual demand of the vegetable crop due to reliance on human experience, dynamically responding to environmental changes by judging whether to execute the light compensation strategy according to the solar elevation angle, determining the backtracking light compensation strategy or the initial light compensation strategy according to the number of occurrences of the growth state in the historical light compensation database, and determining the target light compensation data by reusing mature experience or with the help of a recurrent neural network model, which can not only ensure the stability of light compensation under normal growth state, but also adapt to special or new growth state, ensuring the flexibility and adaptability of light compensation. The historical light compensation records of the light compensation lamp are collected to determine the light compensation swing factor, forming a closed-loop feedback of growth state monitoring, light compensation execution, historical record analysis and parameter adjustment, dynamically controlling the light compensation effect according to the growth state of the vegetable crop and the fluctuation amplitude of the light compensation, and improving the reliability of the light compensation, providing stable photosynthetically active radiation and light duration for the vegetable crop.

[0085] In another preferred mode based on the above embodiments, referring to Figure 2 The present embodiment provides an automatic light compensation control system for vegetable cultivation for applying the above automatic light compensation control method for vegetable cultivation, which comprises:

[0086] The acquisition and judgment module is configured to obtain the crop image information of the vegetable crop, determine the growth state of the vegetable crop based on the crop image information and the crop growth model, and judge whether to execute the light compensation strategy according to the solar elevation angle;

[0087] The strategy determination module is configured to, when it is determined to execute the light compensation strategy, search in the historical light compensation database based on the growth state, and determine the backtracking light compensation strategy or the initial light compensation strategy according to the number of occurrences of the growth state in the historical light compensation database;

[0088] The light supplement determination module is configured to determine corresponding light supplement data in the historical light supplement database based on the growth state when it is determined to be the backtracking light supplement strategy, and determine the target light supplement data of the vegetable crop according to the number of the light supplement data, and determine the target light supplement data of the vegetable crop based on the recurrent neural network model when it is determined to be the initial light supplement strategy;

[0089] The light supplement control module is configured to collect historical light supplement records of the light supplement lamp, analyze the historical light supplement records to determine a light supplement swing factor of the light supplement lamp, adjust the target light supplement data based on the light supplement swing factor, and supplement light to the vegetable crop according to the adjusted target light supplement data.

[0090] Those skilled in the art will understand that the embodiments of the present application can be provided as a method, a system or a computer program product. Therefore, the present application can be in the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can be in the form of a computer program product implemented on one or more computer usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer usable program code.

[0091] The present application is described with reference to flowcharts and / or block diagrams of methods, devices (systems) and computer program products according to embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams can be implemented by computer program instructions, and the combination of flows and / or blocks in the flowcharts and / or block diagrams. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device produce a device that implements the functions specified in the flowcharts and / or block diagrams. Figure 1 one or more flows and / or blocks Figure 1 an apparatus that performs the functions specified in one or more flows or blocks.

[0092] These computer program instructions can also be stored in a computer readable storage medium that can guide the computer or other programmable data processing device to work in a specific way, so that the instructions stored in the computer readable storage medium produce a product including instruction apparatus, which implements the functions specified in the flowcharts and / or block diagrams. Figure 1 one or more flows and / or blocks Figure 1 an apparatus that performs the functions specified in one or more flows or blocks.

[0093] These computer program instructions can also be loaded into a computer or other programmable data processing device, so that a series of operation steps are performed on the computer or other programmable device to produce a computer implemented process, so that the instructions executed on the computer or other programmable device provide a product for implementing the functions specified in the flowcharts and / or block diagrams. Figure 1one or more processes and / or blocks Figure 1 the steps of the functions specified in the one or more blocks.

[0094] Finally, it should be noted that the above-mentioned embodiments are merely used to illustrate the technical solutions of the present application, but not to limit it. Although the present application has been described in detail with reference to the above embodiments, those skilled in the art should understand that the specific embodiments of the present application can be modified or equivalent replaced without departing from the spirit and scope of the present application, and any modification or equivalent replacement should be covered in the protection scope of the claims of the present application.

Claims

1. An automated light supplementation control method for vegetable cultivation, characterized by, The method comprises the following steps: acquiring crop image information of a vegetable crop, determining a growth state of the vegetable crop based on the crop image information and a crop growth model, and determining whether to execute a light supplement strategy according to a solar elevation angle; when it is determined to execute the light supplement strategy, searching in a historical light supplement database based on the growth state, and determining a backtracking light supplement strategy or an initial light supplement strategy according to a number of occurrences of the growth state in the historical light supplement database; when it is determined to be the backtracking light supplement strategy, determining corresponding light supplement data of the vegetable crop in the historical light supplement database based on the growth state, and determining target light supplement data of the vegetable crop according to a number of the light supplement data, and when it is determined to be the initial light supplement strategy, determining the target light supplement data of the vegetable crop based on a recurrent neural network model; collecting historical light supplement records of a light supplement lamp, analyzing the historical light supplement records to determine a light supplement swing factor of the light supplement lamp, adjusting the target light supplement data based on the light supplement swing factor, and supplementing light for the vegetable crop according to the adjusted target light supplement data.

2. The automated light supplementation control method for vegetable cultivation of claim 1, wherein, In the step of determining the growth state of the vegetable crop based on the crop image information and the crop growth model, the method comprises the following steps: acquiring a growth data set of the vegetable crop, dividing the growth data set into a training set and a test set, finding model parameters according to a grid search, and establishing a convolutional neural network model based on the model parameters; training the convolutional neural network model based on the training set, substituting the test set into the trained convolutional neural network model to determine a prediction accuracy, and determining the crop growth model based on the prediction accuracy.

3. The automated light supplementation control method for vegetable cultivation of claim 2, wherein, In the step of determining the growth state of the vegetable crop based on the crop image information and the crop growth model, the method further comprises the following steps: when the prediction accuracy is greater than or equal to a prediction accuracy threshold, the current trained convolutional neural network model is determined as the crop growth model; when the prediction accuracy is less than the prediction accuracy threshold, the amplitude of the change of the current trained convolutional neural network model in the gradient direction is reduced, and the training is continued; the crop image information is substituted into the crop growth model to determine the growth state of the vegetable crop.

4. The automated light supplementation control method for vegetable cultivation of claim 3, wherein, In the step of determining whether to execute the light supplement strategy according to the solar elevation angle, the method comprises the following steps: a solar elevation angle threshold is set in advance, and the solar elevation angle threshold and the solar elevation angle are compared; when the solar elevation angle is greater than or equal to the solar elevation angle threshold, it is determined not to execute the light supplement strategy; when the solar elevation angle is less than the solar elevation angle threshold, it is determined to execute the light supplement strategy; the solar elevation angle threshold is determined based on the growth state.

5. The automated light supplementation control method for vegetable cultivation of claim 4, wherein, In the step of determining the backtracking light supplement strategy or the initial light supplement strategy according to the number of occurrences of the growth state in the historical light supplement database, the method comprises the following steps: the historical light supplement database comprises a plurality of historical growth states, a plurality of historical light supplement data, and a number of occurrences of each historical growth state, and each historical growth state corresponds to a historical light supplement data, and the historical light supplement data comprises a historical light supplement intensity and a historical light supplement angle; determining the backtracking light supplement strategy when the growth state has the same historical growth state in the historical light supplement database and the corresponding occurrence frequency is greater than or equal to the occurrence frequency threshold; determining the initial light supplement strategy when the growth state does not have the same historical growth state in the historical light supplement database or when the growth state has the same historical growth state in the historical light supplement database and the corresponding occurrence frequency is less than the occurrence frequency threshold.

6. The automated light supplementation control method for vegetable cultivation of claim 5, wherein, when the growth state is determined in the historical light supplement database, the corresponding light supplement data is determined, and the number of light supplement data is used to determine the target light supplement data of the vegetable crop, comprising: extracting the same historical growth state of the growth state in the historical light supplement database, and determining the mean of the historical light supplement data corresponding to each historical growth state as the target light supplement data of the vegetable crop.

7. The automated light supplementation control method for vegetable cultivation of claim 6, wherein, when the target light supplement data of the vegetable crop is determined based on the recurrent neural network model, comprising: pre-training the recurrent neural network model and determining the solar radiation track based on the solar tracking algorithm; the input layer of the recurrent neural network model is used to receive the solar radiation track and the growth state; the hidden layer of the recurrent neural network model includes a first LSTM layer, a second LSTM layer, a first full connection layer and a second full connection layer; the number of neurons set in the first LSTM layer is 128, the number of neurons set in the second LSTM layer is 64, and the first full connection layer and the second full connection layer are used to compress features; the output layer of the recurrent neural network model includes a light supplement intensity output layer and a light supplement angle output layer; the number of neurons of the light supplement intensity output layer and the light supplement angle output layer is 1; determining the target light supplement data of the vegetable crop based on the light supplement intensity output layer and the light supplement angle output layer.

8. The automated light supplementation control method for vegetable cultivation of claim 7, wherein, when the historical light supplement record of the light supplement lamp is collected and analyzed to determine the light supplement swing factor of the light supplement lamp, comprising: determining the swing behavior and non-swing behavior in the historical light supplement record, and obtaining the swing number of swing behavior and the non-swing number of non-swing behavior; obtaining the number ratio m of the swing number and the non-swing number; pre-setting a first preset light supplement swing factor, a second preset light supplement swing factor and a third preset light supplement swing factor; when m≤1, the first preset light supplement swing factor is determined as the light supplement swing factor of the light supplement lamp; when 1 when 1.5 the first preset light supplement swing factor is less than the second preset light supplement swing factor, and the second preset light supplement swing factor is less than the third preset light supplement swing factor.

9. The method for automated light supplementation control for vegetable cultivation of claim 8, wherein, when the target light supplement data is adjusted based on the light supplement swing factor, comprising: the target light supplement data is in a proportional relationship with the light supplement swing factor.

10. An automated light supplement control system for vegetable cultivation for applying the automated light supplement control method for vegetable cultivation according to any one of claims 1 to 9, characterized in that, comprising: The collection judgment module is configured to acquire crop image information of the vegetable crop, determine a growth state of the vegetable crop based on the crop image information and a crop growth model, and judge whether to execute a light supplement strategy according to a solar elevation angle; The strategy determination module is configured to, when it is judged to execute the light supplement strategy, search in a historical light supplement database based on the growth state, and determine a backtracking light supplement strategy or an initial light supplement strategy according to a number of occurrences of the growth state in the historical light supplement database; The light supplement determination module is configured to, when it is determined to be the backtracking light supplement strategy, determine corresponding light supplement data in the historical light supplement database based on the growth state, and determine target light supplement data of the vegetable crop according to a number of the light supplement data, and, when it is determined to be the initial light supplement strategy, determine the target light supplement data of the vegetable crop based on a recurrent neural network model; The light supplement control module is configured to collect historical light supplement records of a light supplement lamp, analyze the historical light supplement records to determine a light supplement swing factor of the light supplement lamp, adjust the target light supplement data based on the light supplement swing factor, and supplement light for the vegetable crop according to the adjusted target light supplement data.

Images