A sweet potato intelligent water-fertilizer-pesticide integrated device and control method

Abstract

Disclosed is a sweet potato intelligent water-fertilizer-pesticide integrated device and control method. The device comprises: a water tank, a fertilizer tank, a medicine tank, a drip irrigation device and a controller, wherein the controller is used to automatically adjust the ratio of water, fertilizer and medicine and the drip irrigation time according to the growth conditions and environmental factors of the sweet potatoes. In this way, the intelligent management of the sweet potatoes can be realized, the yield and quality of the sweet potatoes can be improved, and the amount of water, fertilizer and medicine can be saved, and environmental pollution can be reduced.

Description

Technical Field

[0001] The present application relates to the field of intelligent agriculture, and more specifically, to an intelligent water-fertilizer-pesticide integrated device and control method for sweet potatoes. Background Art

[0002] Sweet potato is an important food crop with the characteristics of high yield, stress resistance and rich nutrition. The growth of sweet potato requires the coordination of appropriate water, fertilizer and pesticide, but the traditional manual management method has problems such as low efficiency, large dosage, high cost and heavy pollution. Therefore, in the process of sweet potato planting, in order to ensure the normal growth of sweet potato and to improve the efficiency and quality of farming, a sweet potato water, fertilizer and pesticide integrated device has appeared to automatically drip water and fertilizer for sweet potato.

[0003] However, the traditional integrated water, fertilizer and pesticide device for sweet potatoes usually relies on people's experience and intuition to determine the fertilization ratio and drip irrigation time, which lacks scientific basis and easily causes problems of uneven fertilization or excessive fertilization. In addition, the monitoring means and data processing capabilities of traditional devices are limited, and it is impossible to fully and accurately understand the growth environment and status of sweet potatoes, resulting in low accuracy and reliability of fertilization. Moreover, the drip irrigation data of traditional devices requires people to frequently monitor and adjust, which is cumbersome and error-prone, affecting the fertilization effect and production efficiency.

[0004] Therefore, an optimized intelligent water-fertilizer-pesticide integrated device for sweet potatoes is desired. Summary of the invention

[0005] In order to solve the above technical problems, this application is proposed. The embodiment of this application provides a sweet potato intelligent water, fertilizer and pesticide integrated device and control method, which can realize the intelligent management of sweet potatoes, improve the yield and quality of sweet potatoes, and also save the amount of water, fertilizer and pesticide, and reduce environmental pollution.

[0006] According to one aspect of the present application, a sweet potato intelligent water-fertilizer-pesticide integrated device is provided, which includes: a water tank, a fertilizer tank, a medicine tank, a drip irrigation device and a controller, wherein the controller is used to automatically adjust the ratio of water, fertilizer and medicine and the drip irrigation time according to the growth conditions and environmental factors of the sweet potato.

[0007] According to another aspect of the present application, a method for intelligent integrated water, fertilizer and pesticide control of sweet potatoes is provided, comprising:

[0008] The controller is used to automatically adjust the ratio of water, fertilizer and medicine and the drip irrigation time according to the growth of sweet potatoes and environmental factors; wherein, the controller is used to automatically adjust the ratio of water, fertilizer and medicine and the drip irrigation time according to the growth of sweet potatoes and environmental factors, including:

[0009] Acquire a growth status image of the sweet potato captured by a camera;

[0010] Obtain soil moisture values, soil temperature values, light intensity values, nitrogen content, phosphorus content, and potassium content collected by sensors;

[0011] Extracting features from the growth state image using a growth state feature extractor based on a deep neural network model to obtain a growth state feature map;

[0012] Performing self-attention feature enhancement processing on the growth state feature map to obtain an autocorrelation enhanced growth state feature map;

[0013] Arranging the soil moisture value, the soil temperature value, the light intensity value, the nitrogen content, the phosphorus content and the potassium content as an input vector and passing it through a soil parameter encoder based on a fully connected layer to obtain a soil parameter encoding feature vector;

[0014] Performing feature embedded association analysis on the autocorrelation enhanced growth state characteristic graph and the soil parameter encoding feature vector to obtain soil parameter embedded autocorrelation enhanced growth state characteristics; and

[0015] Based on the soil parameter embedded autocorrelation to enhance the growth state characteristics, the recommended water-fertilizer-pesticide ratio and drip irrigation time are determined.

[0016] Compared with the prior art, the present application provides a sweet potato intelligent water-fertilizer-pesticide integrated device and control method, which includes: a water tank, a fertilizer tank, a medicine tank, a drip irrigation device and a controller, wherein the controller is used to automatically adjust the ratio of water, fertilizer and medicine and the drip irrigation time according to the growth conditions and environmental factors of the sweet potatoes. In this way, the intelligent management of sweet potatoes can be achieved, the yield and quality of sweet potatoes can be improved, and the amount of water, fertilizer and medicine can be saved, and environmental pollution can be reduced. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following briefly introduces the drawings required for use in the description of the embodiments. The following drawings are not intentionally scaled to the actual size, and the focus is on illustrating the main purpose of the present application.

[0018] Figure 1 It is a block diagram of the controller in the sweet potato intelligent water-fertilizer-pesticide integrated device according to an embodiment of the present application.

[0019] Figure 2 This is a flow chart of the intelligent water, fertilizer and pesticide integrated control method for sweet potatoes according to an embodiment of the present application.

[0020] Figure 3 It is a schematic diagram of the system architecture of the intelligent water, fertilizer and medicine integrated control method for sweet potatoes according to an embodiment of the present application.

[0021] Figure 4 This is an application scenario diagram of the sweet potato intelligent water-fertilizer-pesticide integrated device according to an embodiment of the present application. DETAILED DESCRIPTION

[0022] The technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, not all embodiments. Based on the embodiments of the present application, all other embodiments obtained by ordinary technicians in this field without creative work also fall within the scope of protection of the present application.

[0023] As shown in this application and claims, unless the context clearly indicates an exception, the words "a", "an", "an" and / or "the" do not refer to the singular and may also include the plural. Generally speaking, the terms "include" and "comprise" only indicate the inclusion of the steps and elements that have been clearly identified, and these steps and elements do not constitute an exclusive list. The method or device may also include other steps or elements.

[0024] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, any number of different modules can be used and run on the user terminal and / or server. The modules are only illustrative, and different aspects of the system and method can use different modules.

[0025] Flowcharts are used in the present application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed accurately in order. On the contrary, various steps may be processed in reverse order or simultaneously as required. Meanwhile, other operations may also be added to these processes, or a certain step or several steps of operations may be removed from these processes.

[0026] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0027] In view of the above technical problems, in the technical solution of the present application, a sweet potato intelligent water-fertilizer-pesticide integrated device is proposed, which includes: a water tank, a fertilizer tank, a medicine tank, a drip irrigation device and a controller. Accordingly, in the controller, the ratio of water, fertilizer and medicine and the drip irrigation time can be automatically adjusted according to the growth conditions and environmental factors of the sweet potato.

[0028] Specifically, the technical concept of the present application is to collect the growth status images of sweet potatoes through cameras and the growth environment data of sweet potatoes through sensors, and introduce image processing and data analysis algorithms at the back end to conduct comprehensive analysis of these data, so as to determine the recommended water, fertilizer and medicine ratio and drip irrigation time based on the growth conditions and environmental factors of sweet potatoes, thereby realizing intelligent management of sweet potatoes, improving the yield and quality of sweet potatoes, and also saving the use of water, fertilizer and medicine and reducing environmental pollution.

[0029] Figure 1 FIG. 1 is a block diagram of the controller in the sweet potato intelligent water-fertilizer-pesticide integrated device according to an embodiment of the present application. Figure 1 As shown, according to the sweet potato intelligent water-fertilizer-pesticide integrated device of the embodiment of the present application, the controller 100 includes: a sweet potato growth state image acquisition module 110, used to obtain the growth state image of the sweet potato collected by the camera; an environmental data acquisition module 120, used to obtain the soil moisture value, soil temperature value, light intensity value, nitrogen content, phosphorus content and potassium content collected by the sensor; a sweet potato growth state feature extraction module 130, used to extract features from the growth state image through a growth state feature extractor based on a deep neural network model to obtain a growth state feature map; a growth state feature self-attention enhancement module 140, used to perform self-attention feature enhancement processing on the growth state feature map to obtain a self-attention feature map. an environmental data association coding module 150, for arranging the soil moisture value, the soil temperature value, the light intensity value, the nitrogen content, the phosphorus content and the potassium content as an input vector and then passing through a soil parameter encoder based on a fully connected layer to obtain a soil parameter encoding feature vector; a feature embedded association analysis module 160, for performing feature embedded association analysis on the autocorrelation enhanced growth state feature map and the soil parameter encoding feature vector to obtain soil parameter embedded autocorrelation enhanced growth state characteristics; and a water, fertilizer and medicine drip irrigation control module 170, for determining the recommended water, fertilizer and medicine ratio and drip irrigation time based on the soil parameter embedded autocorrelation enhanced growth state characteristics.

[0030] More specifically, in the technical solution of the present application, first, a growth state image of a sweet potato collected by a camera is obtained, and soil moisture values, soil temperature values, light intensity values, nitrogen content, phosphorus content, and potassium content collected by a sensor are obtained. Next, a growth state feature extractor based on a convolutional neural network model with excellent performance in implicit feature extraction of an image is used to perform feature mining on the growth state image to extract implicit feature distribution information about the growth state of the sweet potato in the growth state image, thereby obtaining a growth state feature map.

[0031] Correspondingly, in the sweet potato growth state feature extraction module 130, the deep neural network model is a convolutional neural network model. It is worth mentioning that the convolutional neural network (CNN) is a deep learning model specifically used to process data with a similar grid structure. It has been widely used in the field of computer vision for tasks such as image recognition, object detection, and image segmentation. The core component of the convolutional neural network is the convolutional layer, which can automatically extract features in the image, such as edges, textures, and shapes. By stacking multiple convolutional layers, pooling layers, and fully connected layers, the convolutional neural network can learn high-level features in the image and classify and identify these features. The main advantages of the convolutional neural network are: 1. Parameter sharing: Through the convolution operation, the network can use relatively few parameters to process large-scale image data, reducing the complexity of the model. 2. Translation invariance: The convolutional neural network can identify features in the image, regardless of the location of these features in the image, which makes the convolutional neural network have a certain robustness to the transformation and distortion of the image. 3. Hierarchical feature learning: Through multi-layer convolution and pooling operations, the convolutional neural network can gradually learn the abstract features of the image, from edges and textures to more advanced semantic features. Using the convolutional neural network model in the sweet potato growth status feature extraction module can help extract features in the sweet potato growth status image, such as leaf shape, color, texture, etc., thereby providing richer information for subsequent growth status analysis. These features can be used to determine the health status and growth stage of sweet potatoes, and even predict yields.

[0032] In particular, considering that when the growth state monitoring of sweet potato is actually carried out to control the drip irrigation of water and fertilizer, it is necessary to pay more attention to the state characteristics of the new leaves and roots of sweet potato and the parts related to the growth state change of sweet potato. Therefore, in the technical scheme of the present application, the growth state characteristic graph is further enhanced by a feature-enhanced expressor based on a self-attention mechanism to obtain an autocorrelation-enhanced growth state characteristic graph. It should be understood that the self-attention mechanism is a mechanism that can give different weights according to different parts of the input. Through the application and processing of the self-attention mechanism, the model can pay more attention to the parts related to the growth state change of sweet potato, such as the newly grown leaves, the root state, etc., when processing the growth state characteristic graph of sweet potato, so as to better capture the changes and characteristics of the growth state of sweet potato, which is conducive to better reflecting the growth state information of sweet potato.

[0033] Correspondingly, the growth state feature self-attention enhancement module 140 is used to: pass the growth state feature map through a feature enhancement expressor based on a self-attention mechanism to obtain the autocorrelation enhanced growth state feature map.

[0034] It is worth mentioning that the self-attention mechanism is a mechanism for processing sequence data. It can establish global dependencies in the sequence, so as to give weighted attention to different parts of the sequence in order to better capture the long-distance dependencies in the sequence. The core idea of ​​the self-attention mechanism is to calculate the weight of each position based on the information at different positions in the sequence, and then use these weights to weightedly combine the information at different positions. This mechanism allows the model to give different importance to information at different positions when processing sequence data, so as to better understand and utilize the long-distance dependencies in the sequence. The uses of the self-attention mechanism include but are not limited to the following aspects: 1. Sequence modeling: In natural language processing, the self-attention mechanism can be used to model the dependencies between different words in a sentence, so as to achieve better semantic understanding and generation. 2. Key information extraction: Through the self-attention mechanism, the model can automatically learn and focus on the most important part of the input sequence, so as to achieve the extraction and utilization of key information. 3. Long-distance dependency modeling: The self-attention mechanism helps the model better capture the dependencies between distant positions in the sequence when processing long sequences, such as the association between words in a long sentence. Using the self-attention mechanism in the growth status feature self-attention enhancement module can help the model to pay weighted attention to different parts of the growth status feature map, so as to better capture the dependencies between different positions in the growth status feature map and obtain a growth status feature map with autocorrelation enhancement. This helps to improve the model's understanding and utilization of growth status features, thereby more accurately analyzing and predicting the growth status of sweet potatoes.

[0035] Then, considering that the growth environment parameter data of sweet potatoes, including information such as humidity, temperature, light intensity and nutrient content, have a mutual correlation, in order to more fully dig out the correlation between the various data items in the growth environment parameter data, so as to more accurately control the drip irrigation parameters of water and fertilizer, in the technical solution of the present application, the soil moisture value, the soil temperature value, the light intensity value, the nitrogen content, the phosphorus content and the potassium content are arranged as input vectors and then passed through a soil parameter encoder based on a fully connected layer to obtain a soil parameter encoding feature vector. By encoding with a soil parameter encoder based on a fully connected layer, the correlation feature information between the soil moisture value, the soil temperature value, the light intensity value, the nitrogen content, the phosphorus content and the potassium content can be extracted, which helps to improve the expression ability of soil parameter information and provide a more comprehensive and accurate data basis for the subsequent intelligent recommendation system.

[0036] Furthermore, in order to effectively integrate and learn the soil parameter association characteristics and the sweet potato growth state characteristics, so as to accurately recommend the water-fertilizer-pesticide ratio and drip irrigation time, in the technical solution of the present application, the autocorrelation-enhanced growth state characteristic map and the soil parameter encoding feature vector are further embedded in the channel embedding learner based on the MetaNet module to obtain the soil parameter embedded autocorrelation-enhanced growth state characteristic map. It should be understood that the MetaNet module is a technology for channel embedding learning, which can help to effectively embed and learn information from different sources, so as to obtain a more comprehensive and rich feature representation. In this scenario, through the processing of the MetaNet module, the soil parameter encoding feature vector can be embedded in the high-dimensional space of the autocorrelation-enhanced growth state characteristic map, so as to effectively embed and integrate the sweet potato growth state characteristic information and the sweet potato growth environment association characteristic information, so as to facilitate the subsequent intelligent recommendation system to accurately recommend the water-fertilizer-pesticide ratio and drip irrigation time.

[0037] Correspondingly, the feature embedded association analysis module 160 is used to: embed the autocorrelation enhanced growth state feature map and the soil parameter encoding feature vector through a channel embedding learner based on the MetaNet module to obtain a soil parameter embedded autocorrelation enhanced growth state feature map as the soil parameter embedded autocorrelation enhanced growth state feature.

[0038] Specifically, in one example, the feature embedded association analysis module 160 is used to: embed the autocorrelation enhanced growth state feature map into the convolutional neural network of the channel embedding learner based on the MetaNet module to obtain the growth state feature map; embed the soil parameter encoding feature vector into the one-dimensional convolutional layer of the channel embedding learner based on the MetaNet module to obtain the soil parameter feature vector; and, channel-weight the growth state feature map with the soil parameter feature vector to obtain the soil parameter embedded autocorrelation enhanced growth state feature map.

[0039] It should be understood that the one-dimensional convolution layer is a neural network layer commonly used in deep learning. It is used to process one-dimensional sequence data, such as time series data or text data. The one-dimensional convolution layer can extract local features in the input sequence and share weights at different positions to achieve feature extraction. The one-dimensional convolution layer contains multiple one-dimensional convolution kernels, each of which performs a sliding convolution operation on the input sequence. The size of the convolution kernel determines the length of the area covered when it slides on the input sequence. At each position, the convolution kernel performs a convolution operation with the local window of the input sequence to obtain the feature representation of that position. The one-dimensional convolution layer can effectively capture local patterns in the sequence, such as phrase structures in word sequences or local trends in time series. In general, the one-dimensional convolution layer is an important part of processing one-dimensional sequence data in deep learning. It realizes the extraction and learning of local features in the sequence through convolution operations.

[0040] Channel weighting refers to the weighting operation of each channel in the feature map. Channel weighting can be achieved through learned weights, which can adjust the importance of each channel according to the specific attributes of the input data, thereby affecting the final feature representation. Channel weighting of the growth state feature map for the soil parameter feature vector means weighting each channel in the growth state feature map to consider the impact of soil parameters on the growth state. The process of channel weighting generally includes the following steps: the system learns the weight of each channel through training or other means. These weights can be adjusted according to the specific task and the characteristics of the input data; for each channel in a given feature map, it is multiplied by the corresponding weight, and then the results are added to obtain a weighted feature map. The purpose of channel weighting is to adjust the contribution of each channel in the feature map according to the importance of different attributes or features of the input data on the basis of retaining the original features, so as to obtain a feature representation that is more suitable for a specific task. By performing channel weighting of the soil parameter feature vector on the growth state feature map, it is possible to adjust the importance of different channels in the growth state feature map under the influence of specific soil parameters, so as to obtain a feature representation that is more in line with the influence of soil parameters.

[0041] Then, the soil parameters are embedded in the autocorrelation enhanced growth state characteristic map and passed through a decoder to obtain a decoded value, which is used to represent the recommended water-fertilizer-pesticide ratio and drip irrigation time. In other words, the sweet potato growth state characteristic information embedded and fused through the soil parameter association feature is used for decoding regression, so that the water-fertilizer-pesticide ratio and drip irrigation time are automatically adjusted according to the growth conditions and environmental factors of the sweet potatoes, thereby realizing the intelligent management of the sweet potatoes, improving the yield and quality of the sweet potatoes, and saving the amount of water, fertilizer and pesticides, and reducing environmental pollution.

[0042] Correspondingly, the water, fertilizer and medicine drip irrigation control module 170 is used to: embed the soil parameters into the autocorrelation enhanced growth state characteristic diagram through a decoder to obtain a decoded value, and the decoded value is used to represent the recommended water, fertilizer and medicine ratio and drip irrigation time.

[0043] Furthermore, in the technical solution of the present application, the sweet potato intelligent water, fertilizer and pesticide integrated device also includes a training module for training the growth state feature extractor based on the convolutional neural network model, the feature enhancement expressor based on the self-attention mechanism, the soil parameter encoder based on the fully connected layer, the channel embedding learner based on the MetaNet module and the decoder.

[0044] In one example, the training module includes: a training sweet potato growth state image acquisition unit, which is used to obtain a training growth state image of the sweet potato collected by a camera; a training environment data acquisition unit, which is used to obtain a training soil moisture value, a training soil temperature value, a training light intensity value, a training nitrogen content, a training phosphorus content, and a training potassium content collected by a sensor; a training sweet potato growth state feature extraction unit, which is used to extract features from the training growth state image through the growth state feature extractor based on the deep neural network model to obtain a training growth state feature map; a training growth state feature self-attention enhancement unit, which is used to perform self-attention feature enhancement processing on the training growth state feature map to obtain a training autocorrelation enhanced growth state feature map; a training environment data association encoding unit, which is used to arrange the training soil moisture value, the training soil temperature value, the training light intensity value, the training nitrogen content, the training phosphorus content, and the training potassium content as a training input vector and then perform the training on the training growth state feature map based on the full connection. A soil parameter encoder of the layer is used to obtain a training soil parameter encoding feature vector; a training feature embedded association analysis unit is used to pass the training autocorrelation enhanced growth state feature map and the training soil parameter encoding feature vector through the channel embedding learner based on the MetaNet module to obtain a training soil parameter embedded in the autocorrelation enhanced growth state feature map; a training optimization unit is used to optimize the position-by-position feature value of the training soil parameter embedded in the autocorrelation enhanced growth state feature map to obtain an optimized training soil parameter embedded in the autocorrelation enhanced growth state feature map; a training decoding unit is used to pass the optimized training soil parameter embedded in the autocorrelation enhanced growth state feature map through the decoder to obtain a decoding loss function value; and a training unit is used to train the growth state feature extractor based on the convolutional neural network model, the feature enhancement expressor based on the self-attention mechanism, the soil parameter encoder based on the fully connected layer, the channel embedding learner based on the MetaNet module and the decoder based on the decoding loss function value.

[0045] In particular, in the technical solution of the present application, here, each feature matrix of the training autocorrelation enhanced growth state feature map expresses the image semantic features of the training growth state image, and the distribution in the channel dimension follows the self-attention enhancement based on the spatial distribution of the image semantic features between each feature matrix, so that the channel dimension distribution of the training autocorrelation enhanced growth state feature map also conforms to the spatial distribution of the image semantic features, and in addition, the training soil parameter encoding feature vector expresses the parameter numerical association features between each soil parameter. In this way, when the training autocorrelation enhanced growth state feature map and the training soil parameter encoding feature vector are passed through a channel embedding learner based on the MetaNet module, the channel distribution of the training autocorrelation enhanced growth state feature map is constrained by the parameter numerical association feature distribution between soil parameters expressed by the training soil parameter encoding feature vector, which makes the training soil parameter embedded autocorrelation enhanced growth state feature map as a whole follow the cross-modal image semantic feature-numerical association feature mixed distribution. However, considering that the cross-modal image semantic feature-numerical correlation feature mixed distribution difference will bring local feature distribution sparseness to the overall feature representation of the training soil parameter embedded in the autocorrelation enhanced growth state feature map, that is, the distribution of the sparse sub-manifold outside the overall high-dimensional feature manifold, this will make the training soil parameter embedded in the autocorrelation enhanced growth state feature map when the decoder performs decoding probability regression mapping, the convergence of the training soil parameter embedded in the autocorrelation enhanced growth state feature map to the predetermined regression probability representation in the probability space is poor, affecting the accuracy of the decoded value. Therefore, preferably, the training soil parameter embedded in the autocorrelation enhanced growth state feature map is optimized for position-by-position feature value.

[0046] Accordingly, in one example, the training optimization unit is further used to: optimize the position-by-position characteristic values ​​of the training soil parameter embedded autocorrelation enhanced growth state characteristic map using the following optimization formula to obtain the optimized training soil parameter embedded autocorrelation enhanced growth state characteristic map.

[0047]

[0048] Among them, f i is the characteristic value of the training soil parameter embedded in the autocorrelation enhanced growth state characteristic map F, exp(·) represents the exponential operation of the value, and the exponential operation of the value represents the calculation of the natural exponential function value with the value as the power, f' i It is the characteristic value of the optimized training soil parameter embedded in the autocorrelation enhanced growth state characteristic map.

[0049] That is, by processing the sparse distribution in the high-dimensional feature space based on regularization based on re-probability, the natural distribution transfer from the geometric manifold of the training soil parameter embedded in the autocorrelation enhanced growth state characteristic graph F in the high-dimensional feature space to the probability space is activated, thereby improving the convergence of the complex high-dimensional characteristic manifold with high spatial sparsity under the predetermined regression probability distribution by performing a re-probability-based smoothing regularization on the distribution sparse submanifold of the high-dimensional characteristic manifold of the training soil parameter embedded in the autocorrelation enhanced growth state characteristic graph F, thereby improving the accuracy of the decoded value obtained by the decoder of the training soil parameter embedded in the autocorrelation enhanced growth state characteristic graph F. In this way, the recommended water, fertilizer and medicine ratio and drip irrigation time can be determined based on the growth conditions and environmental factors of sweet potatoes, thereby realizing the intelligent management of sweet potatoes, improving the yield and quality of sweet potatoes, and saving the amount of water, fertilizer and medicine and reducing environmental pollution.

[0050] In summary, the sweet potato intelligent water, fertilizer and pesticide integrated device based on the embodiment of the present application is explained, which can realize the intelligent management of sweet potatoes, improve the yield and quality of sweet potatoes, and also save the use of water, fertilizer and pesticides and reduce environmental pollution.

[0051] As described above, the sweet potato intelligent water-fertilizer-pesticide integrated device based on the embodiment of the present application can be implemented in various terminal devices, such as a server having a sweet potato intelligent water-fertilizer-pesticide integrated control algorithm based on the embodiment of the present application. In one example, the sweet potato intelligent water-fertilizer-pesticide integrated device based on the embodiment of the present application can be integrated into the terminal device as a software module and / or a hardware module. For example, the sweet potato intelligent water-fertilizer-pesticide integrated device based on the embodiment of the present application can be a software module in the operating system of the terminal device, or can be an application developed for the terminal device; of course, the sweet potato intelligent water-fertilizer-pesticide integrated device based on the embodiment of the present application can also be one of the many hardware modules of the terminal device.

[0052] Alternatively, in another example, the sweet potato intelligent water, fertilizer and medicine integrated device based on the embodiment of the present application and the terminal device may also be separate devices, and the sweet potato intelligent water, fertilizer and medicine integrated device may be connected to the terminal device via a wired and / or wireless network, and transmit interactive information in accordance with an agreed data format.

[0053] Furthermore, the present application also provides a method for integrating water, fertilizer and medicine for sweet potatoes, which includes: using a controller to automatically adjust the ratio of water, fertilizer and medicine and the drip irrigation time according to the growth conditions and environmental factors of the sweet potatoes. Figure 2 This is a flow chart of the intelligent water, fertilizer and pesticide integrated control method for sweet potatoes according to an embodiment of the present application. Figure 3 Schematic diagram of the system architecture of the sweet potato intelligent water, fertilizer and medicine integrated control method according to the embodiment of the present application. Figure 2 and Figure 3 As shown, according to the sweet potato intelligent water-fertilizer-pesticide integrated control method of the embodiment of the present application, a controller is used to automatically adjust the ratio of water, fertilizer and pesticide and the drip irrigation time according to the growth conditions and environmental factors of the sweet potato, including: S110, obtaining the growth state image of the sweet potato collected by the camera; S120, obtaining the soil moisture value, soil temperature value, light intensity value, nitrogen content, phosphorus content and potassium content collected by the sensor; S130, extracting features from the growth state image through a growth state feature extractor based on a deep neural network model to obtain a growth state feature map; S140, performing self-attention feature enhancement processing on the growth state feature map to obtain an autocorrelation enhanced growth state characteristic diagram; S150, arranging the soil moisture value, the soil temperature value, the light intensity value, the nitrogen content, the phosphorus content and the potassium content as an input vector and then passing through a soil parameter encoder based on a fully connected layer to obtain a soil parameter encoding feature vector; S160, performing feature embedded association analysis on the autocorrelation enhanced growth state characteristic diagram and the soil parameter encoding feature vector to obtain soil parameter embedded autocorrelation enhanced growth state characteristics; and, S170, determining the recommended water-fertilizer-pesticide ratio and drip irrigation time based on the soil parameter embedded autocorrelation enhanced growth state characteristics.

[0054] Here, those skilled in the art can understand that the specific operations of each step in the above-mentioned sweet potato intelligent water, fertilizer and medicine integrated control method have been referenced above. Figure 1 The description of the sweet potato intelligent water-fertilizer-pesticide integrated device has been introduced in detail, and therefore, its repeated description will be omitted.

[0055] Figure 4 FIG. 1 is an application scenario diagram of the sweet potato intelligent water-fertilizer-pesticide integrated device according to an embodiment of the present application. Figure 4 As shown, in this application scenario, first, the growth state image of the sweet potato captured by the camera is obtained (for example, Figure 4 D1 as shown in the figure), as well as soil moisture value, soil temperature value, light intensity value, nitrogen content, phosphorus content and potassium content collected by the sensor (for example, Figure 4 Then, the growth state image, the soil moisture value, the soil temperature value, the light intensity value, the nitrogen content, the phosphorus content and the potassium content are input to a server (for example, Figure 4In S) shown in , the server can use the sweet potato intelligent water, fertilizer and medicine integrated control algorithm to process the growth status image, the soil moisture value, the soil temperature value, the light intensity value, the nitrogen content, the phosphorus content and the potassium content to obtain a decoded value representing the recommended water, fertilizer and medicine ratio and drip irrigation time.

[0056] The present application uses specific words to describe the embodiments of the present application. For example, "first / second embodiment", "one embodiment", and / or "some embodiments" refer to a certain feature, structure or characteristic related to at least one embodiment of the present application. Therefore, it should be emphasized and noted that "one embodiment" or "an embodiment" or "an alternative embodiment" mentioned twice or multiple times in different positions in this specification does not necessarily refer to the same embodiment. In addition, some features, structures or characteristics in one or more embodiments of the present application can be appropriately combined.

[0057] In addition, it will be appreciated by those skilled in the art that various aspects of the present application may be illustrated and described by a number of patentable categories or situations, including any new and useful process, machine, product or combination of substances, or any new and useful improvements thereto. Accordingly, various aspects of the present application may be performed entirely by hardware, entirely by software (including firmware, resident software, microcode, etc.), or by a combination of hardware and software. The above hardware or software may all be referred to as "data blocks", "modules", "engines", "units", "components" or "systems". In addition, various aspects of the present application may be represented as a computer product located in one or more computer-readable media, which includes computer-readable program code.

[0058] Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by those of ordinary skill in the art to which this application belongs. It should also be understood that terms such as those defined in common dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant technology, and should not be interpreted in an idealized or extremely formal sense, unless explicitly defined as such herein.

[0059] The above is an explanation of the present application and should not be considered as a limitation thereof. Although several exemplary embodiments of the present application are described, it will be readily understood by those skilled in the art that many modifications may be made to the exemplary embodiments without departing from the novel teachings and advantages of the present application. Therefore, all of these modifications are intended to be included within the scope of the present application as defined by the claims. It should be understood that the above is an explanation of the present application and should not be considered to be limited to the specific embodiments disclosed, and modifications to the disclosed embodiments and other embodiments are intended to be included within the scope of the appended claims. The present application is defined by the claims and their equivalents.

Claims

1. A sweet potato intelligent water, fertilizer and pesticide integrated device, comprising: A water tank, a fertilizer tank, a medicine tank, a drip irrigation device and a controller, wherein the controller is used to automatically adjust the ratio of water, fertilizer and medicine and the drip irrigation time according to the growth conditions and environmental factors of the sweet potato; The controller comprises: A sweet potato growth status image acquisition module, used to acquire the growth status image of the sweet potato acquired by a camera; An environmental data acquisition module is used to obtain soil moisture values, soil temperature values, light intensity values, nitrogen content, phosphorus content, and potassium content collected by sensors; A sweet potato growth state feature extraction module, used for extracting features from the growth state image through a growth state feature extractor based on a deep neural network model to obtain a growth state feature map; A growth state feature self-attention enhancement module, used for performing self-attention feature enhancement processing on the growth state feature graph to obtain an autocorrelation enhanced growth state feature graph; An environmental data association encoding module, used for arranging the soil moisture value, the soil temperature value, the light intensity value, the nitrogen content, the phosphorus content and the potassium content as an input vector and then passing it through a soil parameter encoder based on a fully connected layer to obtain a soil parameter encoding feature vector; A feature embedded association analysis module, used for performing feature embedded association analysis on the autocorrelation enhanced growth state characteristic graph and the soil parameter encoding feature vector to obtain soil parameter embedded autocorrelation enhanced growth state characteristics; and A water, fertilizer and medicine drip irrigation control module is used to determine the recommended water, fertilizer and medicine ratio and drip irrigation time based on the soil parameter embedded autocorrelation enhanced growth state characteristics; The sweet potato intelligent water, fertilizer and pesticide integrated device also includes a training module for training a growth state feature extractor based on a deep neural network model, a feature enhancement expressor based on a self-attention mechanism, a soil parameter encoder based on a fully connected layer, a channel embedding learner and a decoder based on a MetaNet module; The training module comprises: A training sweet potato growth state image acquisition unit, used to acquire a training growth state image of the sweet potato acquired by a camera; A training environment data acquisition unit, used to obtain training soil moisture values, training soil temperature values, training light intensity values, training nitrogen content, training phosphorus content, and training potassium content collected by sensors; A training sweet potato growth state feature extraction unit is used to extract features from the training growth state image through the growth state feature extractor based on the deep neural network model to obtain a training growth state feature map; A training growth state feature self-attention reinforcement unit, used for performing self-attention feature reinforcement processing on the training growth state feature map to obtain a training autocorrelation reinforcement growth state feature map; A training environment data association encoding unit, used for arranging the training soil moisture value, the training soil temperature value, the training light intensity value, the training nitrogen content, the training phosphorus content and the training potassium content into a training input vector and then passing it through the fully connected layer-based soil parameter encoder to obtain a training soil parameter encoding feature vector; A training feature embedded association analysis unit, used for embedding the training autocorrelation enhanced growth state feature map and the training soil parameter encoding feature vector into a learner through the channel based on the MetaNet module to obtain a training soil parameter embedded autocorrelation enhanced growth state feature map; A training optimization unit, used for optimizing the position-by-position characteristic value of the training soil parameter embedded autocorrelation enhanced growth state characteristic map to obtain an optimized training soil parameter embedded autocorrelation enhanced growth state characteristic map; A training decoding unit, used for embedding the optimized training soil parameters into the autocorrelation enhanced growth state characteristic map through the decoder to obtain a decoding loss function value; and A training unit, for training the growth state feature extractor based on the deep neural network model, the feature enhancement expressor based on the self-attention mechanism, the soil parameter encoder based on the fully connected layer, the channel embedding learner based on the MetaNet module, and the decoder based on the decoding loss function value; The training optimization unit is further used to optimize the position-by-position characteristic value of the training soil parameter embedded autocorrelation enhanced growth state characteristic map using the following optimization formula to obtain the optimized training soil parameter embedded autocorrelation enhanced growth state characteristic map, in, The training soil parameter embedding autocorrelation enhanced growth state characteristic map The characteristic value of represents an exponential operation of a value, wherein the exponential operation of the value represents calculating the value of a natural exponential function with the value as a power, It is the characteristic value of the optimized training soil parameter embedded in the autocorrelation enhanced growth state characteristic map.

2. The sweet potato intelligent water, fertilizer and pesticide integrated device according to claim 1 is characterized in that: The deep neural network model is a convolutional neural network model.

3. The sweet potato intelligent water, fertilizer and pesticide integrated device according to claim 2 is characterized in that: The growth state feature self-attention reinforcement module is used to: The growth state feature map is passed through a feature enhancement expressor based on a self-attention mechanism to obtain the autocorrelation enhanced growth state feature map.

4. The sweet potato intelligent water, fertilizer and pesticide integrated device according to claim 3 is characterized in that: The feature embedded association analysis module is used to: The autocorrelation enhanced growth state characteristic map and the soil parameter encoding feature vector are passed through a channel embedding learner based on a MetaNet module to obtain a soil parameter embedded autocorrelation enhanced growth state characteristic map as the soil parameter embedded autocorrelation enhanced growth state feature.

5. The sweet potato intelligent water, fertilizer and pesticide integrated device according to claim 4 is characterized in that: The feature embedded association analysis module is used to: Embedding the autocorrelation enhanced growth state feature map into a convolutional neural network of a learner through the channel based on the MetaNet module to obtain a growth state feature map; Embedding the soil parameter encoding feature vector into a one-dimensional convolutional layer of a learner through a channel based on the MetaNet module to obtain a soil parameter feature vector; as well as Channel weighting is performed on the growth state characteristic map using the soil parameter characteristic vector to obtain the soil parameter embedded autocorrelation enhanced growth state characteristic map.

6. The sweet potato intelligent water, fertilizer and pesticide integrated device according to claim 5 is characterized in that: The water, fertilizer and medicine drip irrigation control module is used to: The soil parameters are embedded in the autocorrelation enhanced growth state characteristic diagram and passed through a decoder to obtain a decoded value, and the decoded value is used to represent the recommended water-fertilizer-pesticide ratio and drip irrigation time.

7. A method for controlling the integration of sweet potato intelligent water, fertilizer and medicine, used in the sweet potato intelligent water, fertilizer and medicine integration device as claimed in any one of claims 1 to 6, characterized in that: include: The controller is used to automatically adjust the ratio of water, fertilizer and medicine and the drip irrigation time according to the growth of sweet potatoes and environmental factors; wherein, the controller is used to automatically adjust the ratio of water, fertilizer and medicine and the drip irrigation time according to the growth of sweet potatoes and environmental factors, including: Acquire a growth status image of the sweet potato captured by a camera; Obtain soil moisture values, soil temperature values, light intensity values, nitrogen content, phosphorus content, and potassium content collected by sensors; Extracting features from the growth state image using a growth state feature extractor based on a deep neural network model to obtain a growth state feature map; Performing self-attention feature enhancement processing on the growth state feature map to obtain an autocorrelation enhanced growth state feature map; Arranging the soil moisture value, the soil temperature value, the light intensity value, the nitrogen content, the phosphorus content and the potassium content as an input vector and passing it through a soil parameter encoder based on a fully connected layer to obtain a soil parameter encoding feature vector; Performing feature embedded association analysis on the autocorrelation enhanced growth state characteristic graph and the soil parameter encoding feature vector to obtain soil parameter embedded autocorrelation enhanced growth state characteristics; and Based on the soil parameter embedded autocorrelation to enhance the growth state characteristics, the recommended water-fertilizer-pesticide ratio and drip irrigation time are determined.

Images