A method and system for precise control of water and fertilizer in seedling cultivation based on artificial intelligence

Abstract

The present invention belongs to the field of data processing technology, and specifically relates to a method and system for precise control of water and fertilizer in seedling cultivation based on artificial intelligence. The method includes: determining the growth significance of seedlings at each stage; aligning environmental data of each dimension according to the difference in trend change moments between different dimensions; determining the degree of abnormality of each environmental data in the target dimension based on the degree of influence of each environmental data in the target dimension on the correlation between the target dimension and other dimensions; determining the necessity of correcting each environmental data based on the growth significance of the period in which each environmental data in the target dimension is located and the degree of abnormality of each environmental data; screening the data to be corrected for correction based on the degree of correction necessity; and determining the water and fertilizer requirements of crops based on the corrected environmental data of each dimension. The present invention improves the prediction accuracy of water and fertilizer requirements, thereby promoting the healthy growth of crops.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology. More specifically, the present invention relates to a method and system for precise control of water and fertilizer in seedling cultivation based on artificial intelligence. Background Art

[0002] With the continued growth of the global population and increasing agricultural demands, traditional agricultural production models face a series of challenges, including resource waste, environmental pollution, and the uncertainty brought about by climate change. This necessitates that agricultural production develop in a more scientific, refined, and sustainable direction. Precise control of water and fertilizer management is particularly crucial for plant growth and development, especially during the seedling stage.

[0003] Traditional water and fertilizer management methods often rely on experience and extensive management models, resulting in resource waste and reduced soil fertility. For example, excessive fertilization not only has a negative impact on crops but can also lead to environmental problems such as soil acidification and eutrophication.

[0004] In modern agriculture, sensors and IoT technologies enable real-time monitoring of environmental data during seedling cultivation. Based on changes in this data, the water and fertilizer requirements of seedlings can be predicted, enabling automated irrigation. However, the environmental data collected by sensors can be subject to interference from electromagnetic fields, weather conditions, and other factors, leading to noise. This can affect the accuracy of predictions of seedling water and fertilizer requirements, and consequently, crop growth and yield. Summary of the Invention

[0005] In order to solve the technical problem that noise in the above-mentioned environmental data affects the accuracy of the prediction of the water and fertilizer requirements of the seedlings, the present invention provides solutions in the following aspects.

[0006] In a first aspect, the present invention provides a method for precise control of water and fertilizer in seedling cultivation based on artificial intelligence, comprising:

[0007] Determine the growth significance of seedlings at different periods based on the growth conditions of seedlings; collect environmental data in multiple dimensions; determine multiple trend change moments of environmental data in each dimension; align the environmental data in each dimension according to the trend change moments based on the differences between different dimensions; take any dimension as the target dimension, and determine the degree of abnormality of each environmental data in the target dimension based on the degree of influence of each environmental data in the target dimension on the correlation between the target dimension and other dimensions; determine the necessity of correcting each environmental data based on the growth significance of each environmental data in the target dimension and the degree of abnormality of each environmental data; screen the data to be corrected based on the necessity of correction, and correct the data to be corrected; determine the water and fertilizer requirements of crops based on the corrected environmental data of each dimension.

[0008] Preferably, determining the growth significance of seedlings in each period based on the growth conditions of seedlings includes: obtaining the growth amount of seedlings in each period, and taking the ratio of the average growth amount of all seedlings in the same period to the maximum growth amount of all seedlings in all periods as the growth significance of the corresponding period.

[0009] Preferably, the determination of multiple trend change moments of environmental data of each dimension includes: for any dimension, smoothing the data curve of the dimension, obtaining the moments corresponding to all maximum values and all minimum values in the smoothed data curve, respectively as a trend change moment of the dimension.

[0010] Preferably, the method aligns the environmental data of each dimension in pairs according to the trend change moments based on the difference in trend change moments between different dimensions, including: for any two dimensions, taking the average time difference between each trend change moment in the first dimension and the most recent trend change moment in the second dimension as the misalignment length, recorded as L; in response to the misalignment length being a positive number, removing the first L environmental data in the environmental data of the first dimension and removing the last L environmental data in the environmental data of the second dimension; in response to the misalignment length being a negative number, removing the last L environmental data in the environmental data of the first dimension. The environmental data of the second dimension are removed, and the front The environmental data of each dimension is eliminated to achieve the alignment of the environmental data of these two dimensions.

[0011] Preferably, determining the degree of abnormality of each environmental data in the target dimension includes: obtaining the correlation after alignment between the target dimension and the environmental data of the remaining dimensions as the first correlation; taking any one environmental data in the target dimension as the target data, eliminating the target data in the target dimension, eliminating the environmental data corresponding to the target data in the remaining dimensions, and calculating the correlation between the target dimension and the environmental data of the remaining dimensions after elimination as the second correlation; taking the difference between the first correlation and the second correlation as the degree of influence of the target data on the correlation between the target dimension and the remaining dimensions; obtaining the weighted average of the degree of influence of the target data on the correlation between the target dimension and the remaining dimensions as the degree of abnormality of the target data.

[0012] Preferably, when obtaining the weighted average of the degree of influence of the target data on the correlation between the target dimension and the remaining dimensions, the method for obtaining the weights of the remaining dimensions is: in the data sequence after the target dimension and the remaining dimensions, the absolute value of the difference between the environmental data corresponding to each trend change moment and the environmental data corresponding to the previous trend change moment is used to constitute the trend change amount sequence of the target dimension; in the data sequence after the remaining dimensions and the target dimension, the absolute value of the difference between the environmental data corresponding to each trend change moment and the environmental data corresponding to the previous trend change moment is used to constitute the trend change amount sequence of the remaining dimensions; obtain the DTW distance between the trend change amount sequence of the target dimension and the trend change amount sequence of the remaining dimensions, perform negative correlation normalization on the DTW distance, and use the result of negative correlation normalization as the weight of the remaining dimensions.

[0013] Preferably, the necessity of correction satisfies the expression: ;in, Indicates the necessity of correcting the k-th environmental data in the target dimension; Indicates the abnormality degree of the kth environmental data in the target dimension; represents the growth significance of the period of the kth environmental data in the target dimension; exp( ) is an exponential function with the natural exponential as the base.

[0014] Preferably, the correction of the data to be corrected includes: performing polynomial fitting on local data of the data to be corrected, and using the fitting value of the data to be corrected in the fitting result as the correction value of the data to be corrected; wherein, when performing the polynomial fitting, the fitting weight of each local data is set according to the abnormality degree of each local data, and the fitting weight of each local data is negatively correlated with the abnormality degree of each local data.

[0015] Preferably, determining the water and fertilizer requirements of crops based on the corrected environmental data of each dimension includes: replacing the data to be corrected in the environmental data of each dimension with the corrected environmental data to update the environmental data of each dimension, inputting the updated environmental data of each dimension into a trained neural network, and outputting the water and fertilizer requirements of crops.

[0016] In the second aspect, the present invention provides a seedling water and fertilizer precision control system based on artificial intelligence, including a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the above-mentioned seedling water and fertilizer precision control method based on artificial intelligence is implemented.

[0017] By adopting the above technical solution, the above-mentioned artificial intelligence-based precise control method of water and fertilizer for seedling cultivation is generated into a computer program and stored in a memory to be loaded and executed by a processor, thereby making a terminal device based on the memory and processor for easy use.

[0018] The beneficial effects of the present invention are: the present invention aligns the environmental data of each dimension according to the difference in the trend change moments between different dimensions, thereby avoiding the interference of the delayed changes of the environmental data of each dimension on the correlation analysis between different dimensions, and the correlation analysis of the trend changes of environmental data between different dimensions is more accurate; the present invention determines the abnormality degree of each environmental data according to the influence degree of each environmental data in the target dimension on the correlation between the target dimension and the remaining dimensions, and can accurately distinguish between noise and normal data; the present invention determines the necessity of correction of each environmental data according to the growth significance of the period in which each environmental data in the target dimension is located and the abnormality degree of each environmental data, and selects the correction data according to the necessity of correction for correction, thereby improving the accuracy of the environmental data in the important stage of seedling growth, providing a data basis for crop water and fertilizer demand prediction, and improving the prediction accuracy of water and fertilizer demand, thereby ensuring the healthy growth of crops. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 Schematically illustrates a flow chart of a method for precise control of water and fertilizer in seedling cultivation based on artificial intelligence in the present invention;

[0020] Figure 2 It is a flow chart schematically showing step S3 of a method for precise control of water and fertilizer in seedling cultivation based on artificial intelligence in the present invention. DETAILED DESCRIPTION

[0021] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work shall fall within the scope of protection of the present invention.

[0022] The specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0023] The embodiment of the present invention discloses a method for precise control of water and fertilizer in seedling cultivation based on artificial intelligence, referring to Figure 1 , including steps S1 to S6:

[0024] S1. Determine the growth significance of seedlings at different stages based on the growth of seedlings.

[0025] It should be noted that the growth of seedlings varies at different stages. For example, due to factors such as light, the growth rate of seedlings during the day and at night is different. Therefore, the present invention analyzes the growth of seedlings at different stages to determine the growth significance of seedlings at each stage.

[0026] Specifically, each period, T hours, was defined as the period from seed sowing to the end of the seedling stage and the beginning of the growing season. The growth of the seedlings during each period was calculated, and the ratio of the average growth of all seedlings in the same period to the maximum growth of all seedlings in all periods was used as the growth significance for that period.

[0027] Implementers can set the value of T according to actual implementation conditions, for example, T=6.

[0028] In one embodiment, the method for obtaining the growth amount of the seedlings in each period is as follows: after the seeds germinate, the staff measures the height of the seedlings every T hours, and uses the difference between the current measurement result and the previous measurement result as the growth amount of the seedlings in the corresponding period.

[0029] In another embodiment, the growth of seedlings during each period is determined by placing a camera on the side of a nursery or hotbed. The camera is used to horizontally time-lapse record the entire growth process of the seedlings, starting after sowing and ending at the end of the seedling stage and entering the growth period. The captured video is motion-filtered, and the height of the seedlings at the beginning and end of each period is determined based on the motion-filtered video. The height difference between the beginning and end of each period is used as the growth of the seedlings during each period.

[0030] It should be noted that the present invention performs motion filtering on the seedling growth video, which can eliminate short-term leaf jitter in the video and highlight the growth process of the seedling, making the seedling height obtained based on the video more accurate.

[0031] S2. Collect environmental data in multiple dimensions.

[0032] Environmental data in the seedling environment, including ambient temperature, ambient humidity, light intensity, soil moisture, and soil temperature, are collected in real time using ambient temperature and humidity sensors, light intensity sensors, and soil temperature and humidity sensors deployed in the seedling environment.

[0033] The collection frequency in the present invention is once every 30 minutes, and real-time personnel can set the collection frequency according to actual implementation conditions.

[0034] S3. Determine the degree of abnormality of each environmental data in each dimension.

[0035] See the flowchart of step S3 Figure 2 , including steps S301 to S303, specifically:

[0036] S301: Determine multiple trend change moments of environmental data in each dimension.

[0037] Specifically, for any dimension, a data curve is plotted with time as the horizontal axis and the environmental data of that dimension as the vertical axis. The data curve is smoothed, and the times corresponding to all maximum and minimum values in the smoothed data curve are obtained as the trend change moments of that dimension.

[0038] S302: Based on the differences in trend change times between different dimensions, align the environmental data of each dimension in pairs according to the trend change times.

[0039] It should be noted that the environmental data of each dimension affect each other. For example, when the light intensity increases, the ambient temperature and soil temperature increase, and the ambient humidity and soil humidity decrease. Therefore, changes in environmental data in one dimension may cause changes in environmental data in other dimensions, but this change is delayed, and the degree of delay in changes in environmental data in different dimensions is different. For example, when the light intensity increases, the ambient temperature at the next moment is affected by the light intensity and increases, but after multiple moments, the soil temperature will be affected by the light intensity and increase. The trend change moment of each dimension obtained in step S301 is the moment when the environmental data of each dimension is affected by the environmental data of other dimensions. The present invention aligns the environmental data of different dimensions according to the differences in the trend change moments between different dimensions.

[0040] Specifically, for any two dimensions, the average time difference between each trend change moment in the first dimension and the most recent trend change moment in the second dimension is taken as the dislocation length, denoted as L. For example, when the 1st, 5th, 15th, and 20th moments in the first dimension are trend change moments, and the 2nd, 6th, 16th, and 21st moments in the second dimension are trend change moments, the dislocation length is -1.

[0041] When the offset length is a positive number, the first L environmental data in the first dimension of the environmental data are removed, and the last L environmental data in the second dimension of the environmental data are removed. Conversely, when the offset length is a negative number, the last L environmental data in the first dimension of the environmental data are removed. The environmental data of the second dimension are removed, and the front The environmental data of each dimension is eliminated to achieve the alignment of the environmental data of these two dimensions.

[0042] By performing the above operations on all dimensions pairwise, all dimensions can be aligned.

[0043] S303: Take any dimension as the target dimension, and determine the abnormality of each environmental data in the target dimension according to the influence of each environmental data in the target dimension on the correlation between the target dimension and other dimensions.

[0044] It should be noted that the environmental data of different dimensions are affected by the environmental data of the same dimension and the degree of delay in change is different, but the length of their change time is similar, and the change trends are the same or opposite. Therefore, when the environmental data of two dimensions are aligned, the environmental data of these two dimensions are positively correlated or negatively correlated. When there is noise in the environmental data of one dimension, the correlation between the environmental data of these two dimensions will be deteriorated. Therefore, the present invention eliminates the corresponding environmental data in two dimensions, and determines the degree of influence of the eliminated environmental data on the correlation between the two dimensions by comparing the changes in the correlation between the two dimensions before and after the elimination, thereby obtaining the degree of abnormality of the eliminated environmental data.

[0045] Specifically, the correlation between the target dimension and the environmental data of the remaining dimensions after alignment is calculated as the first correlation.

[0046] Take any environmental data in the target dimension as the target data, eliminate the target data in the target dimension, eliminate the environmental data corresponding to the target data in the remaining dimensions, and calculate the correlation between the target dimension and the environmental data of the remaining dimensions after elimination as the second correlation.

[0047] Implementers can set the correlation algorithm according to actual implementation conditions, such as the cosine similarity algorithm.

[0048] Furthermore, the abnormality of the target data satisfies the expression:

[0049] ;

[0050] Among them, p represents the abnormality degree of the target data in the target dimension; Represents the weight of the hth dimension outside the target dimension; represents the first correlation between the target dimension and the hth dimension outside the target dimension; represents the second correlation between the target dimension and the hth dimension outside the target dimension; N represents the number of dimensions; Indicates the absolute value symbol; max() is the maximum value function, which is used to prevent the abnormal degree of the target data from being negative. The closer it is to 1 or -1, The larger the value, the target dimension is positively or negatively correlated with the hth dimension outside the target dimension. When it is greater than 0, it means that after excluding the target data and the data corresponding to the target data, the target dimension is more correlated with the hth dimension outside the target dimension, which means that the target data affects the correlation between the target dimension and the hth dimension outside the target dimension. The larger it is, the greater the impact of the target data on the correlation between the target dimension and the hth dimension outside the target dimension, the less the target data conforms to the data relationship between these two dimensions, and the more likely the target data is noise; the greater the impact of the target data on the correlation between the target dimension and all other dimensions, the more abnormal the target data is.

[0051] Among them, the weight of the hth dimension outside the target dimension satisfies the expression:

[0052] ;

[0053] in, Represents the weight of the hth dimension outside the target dimension; D represents the trend change sequence of the target dimension; Represents the trend change sequence of the hth dimension outside the target dimension; represents the DTW distance between the trend change sequence of the target dimension and the trend change sequence of the hth dimension outside the target dimension; N represents the number of dimensions; exp( ) is an exponential function with natural exponential as the base, which is used to Perform negative correlation mapping, Used for Perform normalization.

[0054] Among them, the method for obtaining the trend change sequence is:

[0055] In the data sequence after the target dimension is aligned with the remaining dimensions, the absolute value of the difference between the environmental data corresponding to each trend change moment and the environmental data corresponding to the previous trend change moment constitutes the trend change amount sequence of the target dimension; in the data sequence after the remaining dimensions are aligned with the target dimension, the absolute value of the difference between the environmental data corresponding to each trend change moment and the environmental data corresponding to the previous trend change moment constitutes the trend change amount sequence of the remaining dimensions.

[0056] It should be noted that changes in environmental data of the same dimension have different degrees of impact on environmental data of different dimensions. For example, an increase in light intensity may cause the ambient temperature to rise more, for example, the ambient temperature in winter rises from -5°C to 7°C, resulting in a smaller increase in soil temperature, for example, from 0°C to 5°C. The elements in the trend change sequence reflect the size of the local variation range of the environmental data of the corresponding dimension. When the variation ranges of the two dimensions are more similar, the more accurate it is to judge whether the target data is abnormal based on the change in the correlation between the two dimensions before and after the target data is eliminated. Therefore, the present invention determines the weight of the hth dimension outside the target dimension based on the DTW distance between the trend change sequence of the target dimension and the trend change sequence of the hth dimension outside the target dimension. , when the DTW distance between the trend change sequence of the target dimension and the trend change sequence of the h-th dimension is smaller, it means that the data change range of the target dimension and the h-th dimension is more similar, and the weight of the h-th dimension outside the target dimension is greater.

[0057] S4. Determine the necessity of correcting each environmental data based on the growth significance of the period of each environmental data in the target dimension and the degree of abnormality of each environmental data.

[0058] It should be noted that the growth significance of the seedlings in each period reflects the criticality of each growth period of the seedlings. The greater the growth significance, the more the corresponding period is the critical stage of the seedling growth. At this time, the control of water and fertilizer needs to be more accurate to avoid too little or too much water and fertilizer affecting the growth of the seedlings. Therefore, in the critical stage of seedling growth, the more accurate the environmental data of each dimension is required. The abnormality of the environmental data reflects the accuracy of the environmental data. When the abnormality is greater, it means that the environmental data is less accurate due to the influence of noise, and the more it is necessary to correct the environmental data. Therefore, the present invention determines the necessity of correction of each environmental data according to the growth significance of the period in which each environmental data is located in the target dimension and the abnormality of each environmental data, so that the environmental data that need to be corrected can be subsequently screened and corrected according to the necessity of correction.

[0059] Specifically, the modified necessity satisfies the expression:

[0060] ;

[0061] in, Indicates the necessity of correcting the k-th environmental data in the target dimension; Indicates the abnormality degree of the kth environmental data in the target dimension; Indicates the growth significance of the period of the kth environmental data in the target dimension; exp( ) is an exponential function with the natural exponential as the base. The necessity of correcting the kth environmental data in the target dimension is related to the abnormality of the environmental data. When the growth significance of the period of the kth environmental data in the target dimension is When the value is larger, the accuracy requirement for the environmental data is higher. In this case, the gamma transform is used to convert As abnormality The index of abnormality is expanded. When the growth is significant The larger the index The smaller the value, the greater the degree of expansion of the abnormality, making the first The greater the need for correction of environmental data.

[0062] S5. Filter the data to be corrected based on the necessity of correction, and correct the data to be corrected.

[0063] In response to the correction necessity being greater than a preset correction threshold, the corresponding environmental data is used as data to be corrected, wherein the correction threshold is set by the implementation personnel according to the actual implementation situation, for example, 0.3.

[0064] For the data to be corrected, a total of S environmental data before and after it is used as the local data of the data to be corrected. Here, S is a preset number, which is set by the implementer according to the actual implementation situation, for example, S=6.

[0065] All local data of the corrected data are subjected to polynomial fitting using the least squares method. During the fitting process, the loss is:

[0066] ;

[0067] Among them, ε is the loss in the fitting process; is the abnormality degree of the vth local data to be corrected; represents the vth local data of the data to be corrected; f(v) represents the fitting value of the vth local data of the data to be corrected; S represents the preset number; exp( ) represents the exponential function with the natural constant as the base; Represents the fitting weight of the vth local data of the data to be corrected. The greater the abnormality of the vth local data, the more likely it is noise, and the smaller the fitting weight of the vth local data. When performing polynomial fitting, less attention is paid to the vth local data, thus preventing noise from affecting the fitting results.

[0068] It should be noted that when polynomial fitting is performed in the present invention, the highest order term of the polynomial is cubic, then , where a, b, c, and d are the coefficients of the polynomial. Using the loss ε, we take the partial derivatives of a, b, c, and d, setting each derivative equal to 0. This allows us to solve for the coefficients a, b, c, and d, and thus obtain the specific fitting function. This process is a well-known technique in the least squares method and will not be detailed here.

[0069] The fitting value of the data to be corrected in the fitting function is used as the correction value of the data to be corrected. The data to be corrected in the environmental data of each dimension is replaced with its correction value, thus achieving the correction of the environmental data of each dimension.

[0070] S6. Determine the water and fertilizer requirements of crops based on the corrected environmental data in each dimension.

[0071] The present invention uses a neural network to determine the water and fertilizer needs of crops. The specific content of the neural network is as follows: the neural network adopts a fully connected neural network, the input of the neural network is the environmental data after correction of each dimension, including the ambient temperature, ambient humidity, light intensity, soil temperature, soil moisture, etc. at each moment in a time period, and the output of the neural network is the water and fertilizer needs of the crops. The data set of the neural network is the environmental data of each dimension in the historical period, and the label is the water and fertilizer demand. The label is set by relevant professionals based on professional experience. The loss function of the neural network is the mean square error loss. It should be noted that a time period of the present invention is one day, and the implementer can set it according to the actual implementation situation.

[0072] The corrected environmental data of each dimension within a time period is input into the trained neural network to output the water and fertilizer requirements so that the seedling water and fertilizer control system can accurately irrigate and fertilize according to the water and fertilizer requirements.

[0073] An embodiment of the present invention also discloses an artificial intelligence-based precise control system for water and fertilizer in seedling cultivation, which includes a processor and a memory. The memory stores computer program instructions. When the computer program instructions are executed by the processor, an artificial intelligence-based precise control method for water and fertilizer in seedling cultivation according to the present invention is implemented.

[0074] The above system also includes other components well known to those skilled in the art, such as a communication bus and a communication interface. The configuration and functions of these components are known in the art and will not be described in detail here.

[0075] In the description of this specification, “multiple” or “several” means at least two, such as two, three or more, etc., unless otherwise clearly and specifically defined.

[0076] While this specification has shown and described several embodiments of the present invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Numerous modifications, variations, and alternatives will occur to those skilled in the art without departing from the concept and spirit of the present invention. It should be understood that in practicing the present invention, alternatives to the embodiments of the present invention described herein may be employed.

Claims

1. A method for precise control of water and fertilizer in seedling cultivation based on artificial intelligence, characterized in that: include: Determining the growth significance of the seedlings at each stage based on the growth of the seedlings, including: obtaining the growth amount of the seedlings at each stage, and taking the ratio of the average growth amount of all seedlings in the same stage to the maximum growth amount of all seedlings in all stages as the growth significance of the corresponding stage; Collecting environmental data in multiple dimensions; determining multiple trend change moments of the environmental data in each dimension, including: for any dimension, smoothing the data curve of the dimension, obtaining the moments corresponding to all maximum values and all minimum values in the smoothed data curve, each as a trend change moment of the dimension; According to the difference in trend change moments between different dimensions, the environmental data of each dimension are aligned in pairs according to the trend change moments, including: for any two dimensions, the average time difference between each trend change moment in the first dimension and the most recent trend change moment in the second dimension is taken as the misalignment length, recorded as L; in response to the misalignment length being a positive number, the first L environmental data in the environmental data of the first dimension are removed, and the last L environmental data in the environmental data of the second dimension are removed; in response to the misalignment length being a negative number, the last L environmental data in the environmental data of the first dimension are removed. The environmental data of the second dimension are removed, and the front The environmental data of each dimension is eliminated to achieve the alignment of the environmental data of these two dimensions; Taking any one dimension as the target dimension, and determining the abnormality of each environmental data in the target dimension according to the degree of influence of each environmental data in the target dimension on the correlation between the target dimension and the remaining dimensions, including: obtaining the correlation after alignment of the target dimension with the environmental data of the remaining dimensions as the first correlation; taking any one environmental data in the target dimension as the target data, eliminating the target data in the target dimension, eliminating the environmental data corresponding to the target data in the remaining dimensions, and calculating the correlation between the target dimension and the environmental data of the remaining dimensions after elimination as the second correlation; taking the difference between the first correlation and the second correlation as the degree of influence of the target data on the correlation between the target dimension and the remaining dimensions; obtaining the weighted average of the degree of influence of the target data on the correlation between the target dimension and the remaining dimensions as the abnormality of the target data; Determine the necessity of correcting each environmental data based on the growth significance of each environmental data period in the target dimension and the degree of abnormality of each environmental data; select the data to be corrected based on the degree of correction necessity and correct the data to be corrected; Determine the water and fertilizer requirements of crops based on the corrected environmental data in each dimension.

2. The method for precise control of water and fertilizer for seedling cultivation based on artificial intelligence according to claim 1, characterized in that: When obtaining the weighted average of the impact of the target data on the correlation between the target dimension and the remaining dimensions, the weights of the remaining dimensions are obtained as follows: In the data sequence after the target dimension is aligned with the remaining dimensions, the absolute value of the difference between the environmental data corresponding to each trend change moment and the environmental data corresponding to the previous trend change moment is used to form the trend change amount sequence of the target dimension; in the data sequence after the remaining dimensions are aligned with the target dimension, the absolute value of the difference between the environmental data corresponding to each trend change moment and the environmental data corresponding to the previous trend change moment is used to form the trend change amount sequence of the remaining dimensions; The DTW distance between the trend change amount sequence of the target dimension and the trend change amount sequences of the remaining dimensions is obtained, the DTW distance is negatively normalized, and the result of the negative correlation normalization is used as the weight of the remaining dimensions.

3. The method for precise control of water and fertilizer for seedling cultivation based on artificial intelligence according to claim 1, characterized in that: The necessity of the correction satisfies the expression: ； in, Indicates the necessity of correcting the k-th environmental data in the target dimension; Indicates the abnormality degree of the kth environmental data in the target dimension; represents the growth significance of the period of the kth environmental data in the target dimension; exp( ) is an exponential function with the natural exponential as the base.

4. The method for precise control of water and fertilizer for seedling cultivation based on artificial intelligence according to claim 1, characterized in that: The correcting the data to be corrected includes: A polynomial fitting is performed on the local data of the data to be corrected, and the fitting value of the data to be corrected in the fitting result is used as the correction value of the data to be corrected; wherein, when performing the polynomial fitting, a fitting weight of each local data is set according to the abnormality degree of each local data, and the fitting weight of each local data is negatively correlated with the abnormality degree of each local data.

5. The method for precise control of water and fertilizer for seedling cultivation based on artificial intelligence according to claim 1, characterized in that: The water and fertilizer requirements of crops are determined based on the corrected environmental data in each dimension, including: The data to be corrected in the environmental data of each dimension is replaced with the corrected environmental data to update the environmental data of each dimension. The updated environmental data of each dimension is input into the trained neural network to output the water and fertilizer requirements of the crops.

6. A precise control system for water and fertilizer in seedling cultivation based on artificial intelligence, characterized in that: include: A processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, an artificial intelligence-based precise control method of water and fertilizer for seedling cultivation according to any one of claims 1 to 5 is implemented.

Images