WO2022253057A1 - 一种用于日光温室果菜栽培的智能水分精准灌溉控制系统及方法 - Google Patents

一种用于日光温室果菜栽培的智能水分精准灌溉控制系统及方法 Download PDF

Info

Publication number
WO2022253057A1
WO2022253057A1 PCT/CN2022/094818 CN2022094818W WO2022253057A1 WO 2022253057 A1 WO2022253057 A1 WO 2022253057A1 CN 2022094818 W CN2022094818 W CN 2022094818W WO 2022253057 A1 WO2022253057 A1 WO 2022253057A1
Authority
WO
WIPO (PCT)
Prior art keywords
information
irrigation
water demand
fruit
water
Prior art date
Application number
PCT/CN2022/094818
Other languages
English (en)
French (fr)
Inventor
王君
李衍素
于贤昌
卢威
袁泉
陈茹
孙敏涛
贺超兴
闫妍
Original Assignee
中国农业科学院蔬菜花卉研究所
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 中国农业科学院蔬菜花卉研究所 filed Critical 中国农业科学院蔬菜花卉研究所
Publication of WO2022253057A1 publication Critical patent/WO2022253057A1/zh

Links

Images

Classifications

    • A—HUMAN NECESSITIES
    • A01—AGRICULTURE; FORESTRY; ANIMAL HUSBANDRY; HUNTING; TRAPPING; FISHING
    • A01G—HORTICULTURE; CULTIVATION OF VEGETABLES, FLOWERS, RICE, FRUIT, VINES, HOPS OR SEAWEED; FORESTRY; WATERING
    • A01G9/00—Cultivation in receptacles, forcing-frames or greenhouses; Edging for beds, lawn or the like
    • A01G9/24—Devices or systems for heating, ventilating, regulating temperature, illuminating, or watering, in greenhouses, forcing-frames, or the like
    • A01G9/247—Watering arrangements
    • A—HUMAN NECESSITIES
    • A01—AGRICULTURE; FORESTRY; ANIMAL HUSBANDRY; HUNTING; TRAPPING; FISHING
    • A01G—HORTICULTURE; CULTIVATION OF VEGETABLES, FLOWERS, RICE, FRUIT, VINES, HOPS OR SEAWEED; FORESTRY; WATERING
    • A01G25/00—Watering gardens, fields, sports grounds or the like
    • A01G25/16—Control of watering
    • A01G25/167—Control by humidity of the soil itself or of devices simulating soil or of the atmosphere; Soil humidity sensors
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
    • Y02A40/00—Adaptation technologies in agriculture, forestry, livestock or agroalimentary production
    • Y02A40/10—Adaptation technologies in agriculture, forestry, livestock or agroalimentary production in agriculture
    • Y02A40/22—Improving land use; Improving water use or availability; Controlling erosion
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
    • Y02A40/00—Adaptation technologies in agriculture, forestry, livestock or agroalimentary production
    • Y02A40/10—Adaptation technologies in agriculture, forestry, livestock or agroalimentary production in agriculture
    • Y02A40/25—Greenhouse technology, e.g. cooling systems therefor

Definitions

  • the invention relates to the field of crop irrigation technology, in particular to an intelligent water precision irrigation control system and method for fruit and vegetable cultivation in a solar greenhouse.
  • This application requires the application number submitted to the Patent Office of the State Intellectual Property Office of China on June 1, 2021: CN202110607031.3 and the application number submitted to the Patent Office of the State Intellectual Property Office of China on September 6, 2021: CN202111041109. 6, the entire disclosure of which is incorporated herein by reference in its entirety.
  • the area of horticultural facilities in my country has reached 28.4 million mu, of which solar greenhouses account for about 31%.
  • the types of fruits and vegetables planted mainly include peppers, tomatoes, cucumbers, eggplants, etc.
  • Facility off-season cultivation has become an important part of people's daily fruit and vegetable supply.
  • producers rely on experience to perform extensive irrigation, resulting in waste of water resources, reducing fertilizer resource utilization efficiency and fruit quality. Therefore, realizing precise irrigation based on the law of water demand of fruits and vegetables and the external environment is particularly important for water saving and quality improvement.
  • Chinese patent document CN 109845625 A discloses a neural network-based multi-dimensional parameter intelligent irrigation control method for crops, by collecting parameters such as rainfall information, soil moisture information, wind speed information, temperature and humidity information, light intensity information and flow information of the current irrigated farmland. Based on the neural network, the crop water demand model is established based on the crop water demand signal as the response information. Through this model, the multi-dimensional environmental parameters of the crops are calculated and processed, and finally the water demand of the current farmland crops is predicted. The controller controls the water demand and rainfall. and soil moisture to make a comprehensive judgment result and control the solenoid valve according to the judgment result to realize the irrigation of crops.
  • this technical scheme considers the factors affecting irrigation water demand more comprehensively, the model constructed is very simple, and the accuracy of water demand prediction results is not high.
  • the applicant uses the proposed deep neural network to carry out intelligent water precision irrigation control, which greatly improves the prediction accuracy of water demand.
  • This application realizes the precise irrigation control of various fruits and vegetables in different regions, that is, the same control system can be used for various fruits and vegetables, which improves the versatility.
  • this application adopts new algorithms in the preprocessing, segmentation, pooling layer, new activation function R l (a parallel implementation scheme different from the cosine activation function), and the use of loss functions , to improve the accuracy and speed of deep neural network training as a whole.
  • new activation function R l a parallel implementation scheme different from the cosine activation function
  • loss functions loss functions
  • the G component is enhanced, and the R and B components are relatively suppressed, which is conducive to improving the accuracy of demand forecasting and judgment .
  • the inventor is committed to the prediction of water demand for precise irrigation of fruits and vegetables, and has proposed a number of parallel association schemes, and without causing conflicts, each module can be cross-combined, and all combinations constitute the scope of this application within.
  • the concept of deep learning was proposed by Hinton et al. in 2006. Based on the deep belief network (DBN), a non-supervised greedy layer-by-layer training algorithm is proposed, which brings hope to solve the optimization problems related to the deep structure, and then a multi-layer autoencoder deep structure is proposed.
  • the convolutional neural network proposed by Lecun et al. is the first real multi-layer structure learning algorithm, which uses the spatial relative relationship to reduce the number of parameters to improve training performance.
  • Deep learning is a new field in machine learning research. Its motivation is to establish and simulate the neural network of human brain for analysis and learning. It imitates the mechanism of human brain to explain data, such as images, sounds and texts. Like machine learning methods, deep machine learning methods are also divided into supervised learning and unsupervised learning. The learning models established under different learning frameworks are very different. For example, Convolutional neural networks (CNNs) is a machine learning model under deep supervised learning, and Deep Belief Nets (DBNs) is a machine learning model under unsupervised learning. .
  • CNNs Convolutional neural networks
  • DNNs Deep Belief Nets
  • CNN Convolutional Neural Networks
  • Feedforward Neural Networks Feedforward Neural Networks
  • It is one of the representative algorithms for deep learning.
  • the deep convolutional neural network DCNN is a network structure with multiple CNN layers.
  • the activation functions often used in deep neural networks are as follows: sigmoid function, tanh function, ReLU function.
  • the sigmoid function which maps the value of (- ⁇ ,+ ⁇ ) to (0,1).
  • the formula of the sigmoid function is as follows:
  • the sigmoid function is used as a nonlinear activation function, but it is not often used. It has the following disadvantages:
  • the tanh function is more common than the sigmoid function. This function maps the value of (- ⁇ ,+ ⁇ ) to (-1,1).
  • the formula is:
  • the tanh function can be regarded as linear in a short region around 0. Since the mean value of the tanh function is 0, it makes up for the shortcoming of the mean value of the sigmoid function being 0.5.
  • the ReLU function also known as the Rectified Linear Unit, is a piecewise linear function that makes up for the gradient disappearance problem of the sigmoid function and the tanh function.
  • the formula of the ReLU function is as follows:
  • this application proposes an intelligent water precision irrigation control system for fruit and vegetable cultivation in a solar greenhouse.
  • the system includes:
  • the information collection module collects information through various sensors, and the collected information includes: at least one of soil information, ground environment information, irrigation equipment use process information, and crop information;
  • the water demand prediction module inputs the collected information into the trained deep neural network for calculation, and predicts the current water demand for growing fruits and vegetables in the solar greenhouse;
  • the irrigation control module forms control instructions according to the predicted water demand of the fruits and vegetables, so as to supply water regularly and quantitatively.
  • the soil information includes: at least one item of soil texture, soil field water holding capacity, soil temperature, and soil moisture information.
  • the ground environment information includes: at least one item of the current temperature and humidity information of the air inside the solar greenhouse, the current light intensity of the fruit and vegetable canopy, and the cumulative light radiation information.
  • the irrigation equipment use process information includes: at least one of irrigation pipe diameter information, flow rate information, irrigation duration and irrigation water volume.
  • the crop information includes fruit and vegetable crop species, growth stage information, and growth state information.
  • the sensors used in this application include: soil moisture sensor, light intensity sensor, air temperature and humidity sensor and image sensor.
  • the present application also proposes an intelligent water precision irrigation control method for fruit and vegetable cultivation in a solar greenhouse, the method comprising:
  • Use the water demand prediction module to input the collected information into the trained deep neural network for calculation, and predict the water demand of the current solar greenhouse cultivation of fruits and vegetables;
  • the irrigation control module is used to form control instructions according to the predicted water demand of the fruits and vegetables, so as to perform regular and quantitative water supply.
  • the soil information includes: at least one item of soil texture, soil field water holding capacity, soil temperature, and soil moisture information.
  • the ground environment information includes: at least one item of the current temperature and humidity information of the air inside the solar greenhouse, the current light intensity of the fruit and vegetable canopy, and the cumulative light radiation information.
  • the irrigation equipment use process information includes: at least one of irrigation pipe diameter information, flow rate information, irrigation duration and irrigation water volume.
  • the crop information includes fruit and vegetable crop species, growth stage information, and growth state information.
  • the sensors used in this application include: soil moisture sensor, light intensity sensor, air temperature and humidity sensor and image sensor.
  • the present application also correspondingly proposes a computer storage medium, on which a program code is stored, and the code is used to implement any one of the methods described above.
  • the present application also correspondingly proposes a computer device, the device includes a processor and a memory, and the memory stores computer instructions, and the instructions are used to implement any one of the methods described above.
  • FIG. 1 shows a schematic structural view of a basic embodiment of the present application.
  • the present application proposes an intelligent water precision irrigation control system for fruit and vegetable cultivation in a solar greenhouse.
  • the system includes:
  • the information collection module collects information through various sensors, and the collected information includes: at least one of soil information, ground environment information, irrigation equipment use process information, and crop information;
  • the water demand prediction module inputs the collected information into the trained deep neural network for calculation, and predicts the current water demand for growing fruits and vegetables in the solar greenhouse;
  • the irrigation control module forms control instructions according to the predicted water demand of the fruits and vegetables, so as to supply water regularly and quantitatively.
  • the soil information includes: at least one item of soil texture, soil field water holding capacity, soil temperature, and soil moisture information.
  • the ground environment information includes: at least one item of the current temperature and humidity information of the air inside the solar greenhouse, the current light intensity of the fruit and vegetable canopy, and the cumulative light radiation information.
  • the irrigation equipment use process information includes: at least one of irrigation pipe diameter information, flow rate information, irrigation duration and irrigation water volume.
  • the crop information includes fruit and vegetable crop species, growth stage information, and growth state information.
  • the sensors used in this application include: soil moisture sensors, light intensity sensors, temperature and humidity sensors, and image sensors.
  • the image sensor acquires the image information of the fruit and vegetable, and inputs the acquired image information into the fruit and vegetable recognition model to obtain information on the type and growth stage of the fruit and vegetable.
  • the water demand prediction module utilizes the water demand prediction model to obtain the water demand
  • the water demand prediction model is a deep convolutional neural network DCNN
  • the deep convolutional neural network DCNN may include: an information input layer, one or Multiple convolutional layers, one or more pooling layers, one or more hidden layers, fully connected layers; the convolution kernel size used by the convolutional layer is 3 ⁇ 3; the pooling layer is the largest pool Calculation method;
  • the activation function adopted by the deep convolutional neural network DCNN is a cosine activation function, denoted as f (), where
  • ⁇ yi represents the vector angle between sample i and its corresponding label y i ;
  • N represents the number of training samples;
  • w yi represents the weight of sample i at its label y i ;
  • the information input layer receives information from one or more sensors
  • the fruit and vegetable water demand forecasting model receives output results from the fruit and vegetable identification model, including: at least one item of information on fruit and vegetable types, fruit and vegetable growth stages, and fruit and vegetable growth states.
  • the fruit and vegetable recognition model uses the VGG-16 network model.
  • the training set is used for training.
  • the training set contains a variety of fruits and vegetables, including: cucumber, tomato, pepper, eggplant and more than 100 kinds of fruit and vegetable images, a total of 25,783 images, of which 75% are used as training sets and 25% as test sets.
  • AdamOptimizer is also used as an optimizer, using convolutional layers with neurons of 64 ⁇ 32 and 32 ⁇ 16 respectively, using 3 ⁇ 3 convolution kernels, and a maximum pooling layer with a step size of 2, 64 Fully connected layers of neurons and softmax regression layers are implemented.
  • the optimizer includes: at least one of GradientDescentOptimizer, AdagradOptimizer, AdagradDAOptimizer, MomentumOptimizer, and RMSPropOptimizer.
  • preprocessing the acquired image information of fruits and vegetables also includes: screening of image information, eliminating images that do not meet the definition requirements or fail to contain at least one complete leaf;
  • Image clipping includes: clipping the obtained image information to obtain leaf information.
  • the preprocessing also includes: suppressing the color of the R and B components in the RGB space, and enhancing the G channel component.
  • the G component is enhanced, and the R, G, and B components of the segmented leaves are first separated; the adjustment coefficient of the G component is generated:
  • the decomposition and synthesis of each component belongs to the existing technology in this field, but the above adjustment method is the inventor's first creation. After the above adjustment, the green channel information of the blade can be fully utilized to make the subsequent prediction process more accurate.
  • the water demand prediction model specifically includes a multi-region convolutional neural network model
  • the multi-regional convolutional neural network model includes: a convolutional network layer for generating mapping features of the original blade; multi-region confidence
  • the network model including the confidence network model of multiple regions, is used to generate a plurality of different confidence values of different water demands for the current state of the fruits and vegetables, and fit the different confidence values of multiple regions to determine the If the confidence values in the area are all relatively large, the corresponding water demand is determined as the water demand of fruits and vegetables.
  • the multi-region belief network model includes a multi-region pooling layer and a fully connected layer
  • the multi-region pooling layer includes pooling layers in multiple regions, the number of pooling layers is 1, and the pooling layer is also Can be replaced by a fully connected layer; the pooling layer is used to generate confidence.
  • the image segmentation adopts an improved watershed segmentation method for segmentation:
  • Gradient(x,y) represents the original gradient value of the pixel point (x,y); Respectively represent the average gradient value, minimum gradient value, and maximum gradient value in the window D area; Gra represents the corrected gradient value;
  • the water demand forecasting model includes a convolutional network layer and a multi-area belief network
  • the convolutional network layer is used to generate the features of each segmented leaf
  • the multi-area belief network includes multiple leaf areas
  • the belief network model is used to generate confidence values for multiple different leaf water shortage degrees. When the confidence values of at least two leaf regions meet the preset first threshold level range, the water shortage at this time is used as Water requirements of fruits and vegetables.
  • the water demand forecasting model includes a convolutional network layer and a multi-region belief network, and the convolutional network layer can receive the segmented leaf images of the captured images and generate different scales.
  • Convolution multi-feature map The multi-region belief network model includes a multi-region pooling layer and a fully connected layer; Wherein, the multi-region pooling layer includes a plurality of region pooling layers; the plurality of region pooling layers Used to generate the confidence value of the water demand of the fruit vegetable image, when the confidence value of the confidence value of at least two leaf regions meets the preset first threshold level range, then determine the water demand at this time as the water demand of the fruit vegetable ;
  • the multi-region pooling layer is set as a maximum pooling layer or an average pooling layer; the fully connected layer is used to classify the identified degree of leaf water shortage.
  • the water demand prediction model is a deep convolutional neural network, specifically including: an input layer, an embedding layer, a pooling layer, and a fully connected layer; the input layer is used to receive Segmented leaf image; the convolution kernel size used in the embedding layer is 5 ⁇ 5; the activation function is denoted as R l (); after being processed by the fully connected layer, the water demand prediction recommendation result is further obtained;
  • the pooling method of the pooling layer is as follows:
  • x e represents the output of the current layer
  • ue represents the input of the activation function R l
  • R l () represents the activation function
  • w e represents the weight of the current layer
  • ⁇ represents the loss function
  • x e-1 represents the output of the previous layer
  • the activation function R l is:
  • the loss function ⁇ is as follows:
  • N represents the size of the positive sample data set, i takes a value from 1 to N, y i represents the label value corresponding to the positive sample x i ; W yi represents the weight of the positive sample feature vector x i at its label y i , and s is the depth volume
  • the recommended parameters of the product neural network; b j represents the deviation of the sample xi at its label y i .
  • the present application also proposes an intelligent water precision irrigation control method for fruit and vegetable cultivation in a solar greenhouse, the method comprising:
  • Use the water demand prediction module to input the collected information into the trained deep neural network for calculation, and predict the water demand of the current solar greenhouse cultivation of fruits and vegetables;
  • the irrigation control module is used to form control instructions according to the predicted water demand of the fruits and vegetables, so as to perform regular and quantitative water supply.
  • the soil information includes: at least one item of soil texture, soil field water holding capacity, soil temperature, and soil moisture information.
  • the ground environment information includes: at least one item of the current temperature and humidity information of the air inside the solar greenhouse, the current light intensity of the fruit and vegetable canopy, and the cumulative light radiation information.
  • the irrigation equipment use process information includes: at least one of irrigation pipe diameter information, flow rate information, irrigation duration and irrigation water volume.
  • the crop information includes fruit and vegetable crop species, growth stage information, and growth state information.
  • the sensors used in this application include: soil moisture sensors, light intensity sensors, temperature and humidity sensors, and image sensors.
  • the image sensor acquires the image information of the fruit and vegetable, and inputs the acquired image information into the fruit and vegetable recognition model to obtain information on the type and growth stage of the fruit and vegetable.
  • the water demand forecasting module uses a water demand forecasting model to obtain the water demand
  • the water demand forecasting model is a deep convolutional neural network DCNN
  • the deep convolutional neural network DCNN includes: an information input layer, one or more Convolutional layer, one or more pooling layers, one or more hidden layers, fully connected layer; the convolution kernel size used by the convolutional layer is 3 ⁇ 3; the pooling layer is the maximum pooling method to calculate; the activation function adopted by the deep convolutional neural network DCNN is a cosine activation function, denoted as f (), where
  • the information input layer receives information from one or more sensors
  • the fruit and vegetable water demand forecasting model receives output results from the fruit and vegetable identification model, including: at least one item of information on fruit and vegetable types, fruit and vegetable growth stages, and fruit and vegetable growth states.
  • the fruit and vegetable recognition model uses the VGG-16 network model.
  • the training set contains a variety of fruits and vegetables, including: eggplant, tomato, cucumber, cabbage, potato and more than 100 kinds of fruit and vegetable images, a total of 25783 images, 75% of which are used as training sets and 25% as testing set.
  • an optimizer is also used for optimization, using convolutional layers with neurons of 64 ⁇ 32 and 32 ⁇ 16 respectively, using 3 ⁇ 3 convolution kernels, and a maximum pooling layer with a step size of 2, 64 Fully connected layers of neurons and softmax regression layers are implemented.
  • the optimizer includes: at least one of GradientDescentOptimizer, AdagradOptimizer, AdagradDAOptimizer, MomentumOptimizer, and RMSPropOptimizer.
  • preprocessing the acquired image information of fruits and vegetables also includes: screening of image information, eliminating images that do not meet the definition requirements or fail to contain at least one complete leaf;
  • Image clipping includes: clipping the obtained image information to obtain leaf information.
  • the preprocessing further includes: performing color suppression on the R and B components in the RGB space, and enhancing the G channel component.
  • the G component is enhanced, and the R, G, and B components of the segmented leaves are first separated; the adjustment coefficient of the G component is generated:
  • the decomposition and synthesis of each component belongs to the existing technology in this field, but the above adjustment method is the inventor's first creation. After the above adjustment, the green channel information of the blade can be fully utilized to make the subsequent prediction process more accurate.
  • the water demand prediction model specifically includes a multi-area convolutional neural network model
  • the multi-area convolutional neural network model includes: a convolutional network layer for generating the original blade
  • the mapping feature of the multi-area belief network model including the belief network model of multiple regions, is used to generate a plurality of different confidence values of different water demands for the current state of the fruit and vegetables, and perform different confidence values for multiple regions. Fitting, determine the confidence value with relatively large confidence value in different regions, and determine the corresponding water demand as the water demand of fruits and vegetables.
  • the multi-region belief network model includes a multi-region pooling layer and a fully connected layer
  • the multi-region pooling layer includes pooling layers in multiple regions, the number of pooling layers is 1, and the pooling layer is also Can be replaced by a fully connected layer; the pooling layer is used to generate confidence.
  • the image segmentation adopts an improved watershed segmentation method for segmentation:
  • Gradient(x,y) represents the original gradient value of the pixel point (x,y); Respectively represent the average gradient value, minimum gradient value, and maximum gradient value in the window D area; Gra represents the corrected gradient value;
  • the water demand forecasting model includes a convolutional network layer and a multi-area belief network
  • the convolutional network layer is used to generate the features of each segmented leaf
  • the multi-area belief network includes multiple leaf areas
  • the belief network model is used to generate confidence values for multiple different leaf water shortage degrees. When the confidence values of at least two leaf regions meet the preset first threshold level range, the water shortage at this time is used as Water requirements of fruits and vegetables.
  • the water demand forecasting model includes a convolutional network layer and a multi-region belief network, and the convolutional network layer can receive the segmented leaf images of the captured images and generate different scales.
  • Convolution multi-feature map The multi-region belief network model includes a multi-region pooling layer and a fully connected layer; Wherein, the multi-region pooling layer includes a plurality of region pooling layers; the plurality of region pooling layers Used to generate the confidence value of the water demand of the fruit vegetable image, when the confidence value of the confidence value of at least two leaf regions meets the preset first threshold level range, then determine the water demand at this time as the water demand of the fruit vegetable ;
  • the multi-region pooling layer is set as a maximum pooling layer or an average pooling layer; the fully connected layer is used to classify the identified degree of leaf water shortage.
  • the water demand prediction model is a deep convolutional neural network, specifically including: an input layer, an embedding layer, a pooling layer, and a fully connected layer; the input layer is used to receive Segmented leaf image; the convolution kernel size used in the embedding layer is 5 ⁇ 5; the activation function is denoted as R l (); after being processed by the fully connected layer, the water demand prediction recommendation result is further obtained;
  • the pooling method of the pooling layer is as follows:
  • x e represents the output of the current layer
  • ue represents the input of the activation function R l
  • R l () represents the activation function
  • w e represents the weight of the current layer
  • ⁇ represents the loss function
  • x e-1 represents the output of the previous layer
  • the activation function R l is:
  • the loss function ⁇ is as follows:
  • N represents the size of the positive sample data set, i takes a value from 1 to N, y i represents the label value corresponding to the positive sample x i ; W yi represents the weight of the positive sample feature vector x i at its label y i , and s is the depth volume
  • the recommended parameters of the product neural network; b j represents the deviation of the sample xi at its label y i .
  • the present application also proposes a computer-readable medium, which contains program codes that can realize the above-mentioned system, and the contained program codes can be transmitted by any appropriate medium, including but not limited to wireless, electric wires, optical cables, RF, etc., or the above-mentioned any suitable combination.
  • the computer readable medium may be a computer readable signal medium or a computer readable storage medium.
  • a computer-readable storage medium may be, for example, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or device, or any combination thereof.
  • a computer-readable storage medium may be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.
  • Computer program code for carrying out the operations of the present invention may be written in one or more programming languages, or combinations thereof, including object-oriented programming languages—such as Java, Smalltalk, C++, and conventional Procedural Programming Language - such as "C" or a similar programming language.
  • the program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server.
  • the remote computer can be connected to the user computer through any kind of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computer (such as through the Internet using an Internet service provider). connect).
  • LAN local area network
  • WAN wide area network
  • connect such as AT&T, MCI, Sprint, EarthLink, MSN, GTE, etc.
  • the above-mentioned integrated units realized in the form of software functional units may be stored in a computer-readable storage medium.
  • the above-mentioned software functional units are stored in a storage medium, and include several instructions to enable a computer device (which may be a personal computer, server, or network device, etc.) or a processor (processor) to execute the methods described in various embodiments of the present invention. partial steps.
  • the aforementioned storage media include: U disk, mobile hard disk, read-only memory (Read-Only Memory, ROM), random access memory (Random Access Memory, RAM), magnetic disk or optical disc and other media that can store program codes. .

Landscapes

  • Life Sciences & Earth Sciences (AREA)
  • Environmental Sciences (AREA)
  • Soil Sciences (AREA)
  • Engineering & Computer Science (AREA)
  • Water Supply & Treatment (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

本申请涉及一种用于日光温室果菜栽培的智能水分精准灌溉控制系统及方法，所述系统包括：信息采集模块，通过多种传感器进行信息采集，所采集的信息包括：土壤信息、地上环境信息、灌水设备使用过程信息、作物信息中的至少一项；需水量预测模块，将所采集的信息输入至深度神经网络进行计算，预测出当前日光温室栽培果菜的需水量；灌溉控制模块，根据预测出的所述蔬菜需水量形成控制指令，以进行定时定量的供水。通过上述方式，本发明能够根据蔬菜的生长情况和环境参量决策灌溉时间和灌溉量，全程无需人工操作，能够实现设施栽培智能灌溉，工作效率高，且精确度高，实现蔬菜节水提质增效生产。

Description

一种用于日光温室果菜栽培的智能水分精准灌溉控制系统及方法 技术领域
本发明涉及农作物灌溉技术领域，尤其涉及一种用于日光温室果菜栽培的智能水分精准灌溉控制系统及方法。本申请要求在2021年6月1日向中国国家知识产权局专利局提交的申请号为：CN202110607031.3的申请和2021年9月6日向中国国家知识产权局专利局提交的申请号为：CN202111041109.6的申请为优先权，其整个公开通过引用的方式全部并入此处。
背景技术
我国园艺设施面积已达2840万亩，其中，日光温室约占31％，种植果菜种类主要包含辣椒、番茄、黄瓜、茄子等。设施反季节栽培已成为人们日常果菜供应重要的组成部分。但是，在实际生产过程中，生产者多凭借经验进行粗放灌溉，造成水资源浪费，降低肥料资源利用效率和果实品质。因此，实现基于果菜水分需求规律和的外界环境精准灌溉对于节水提质变得尤为重要。
中国专利文献CN 109845625 A公开了一种基于神经网络的多维参量农作物智能灌溉控制方法，通过采集当前灌溉农田的雨量信息、土壤墒情信息、风速信息、温湿度信息、光照强度信息及流量信息等参量，基于神经网络建立以农作物需水信号为响应信息的农作物需水量模型，通过该模型对农作物多维环境参量进行计算处理，最终预测出当前农田农作物的需水量，控制器通过对需水量、降雨量及土壤墒情做出综合判决结果并根据判决结果控制电磁阀，实现对农作物的灌溉。该技术方案虽然考虑影响灌溉需水量的因素较全面，所构建的模型十分简单，需水量预测结果精度不高。
为解决上述问题，申请人利用所提出的深度神经网络进行智能水分精准灌溉控制，极大提高了需水量的预测精度。
本申请的创造性贡献在于：
1.本申请实现了对不同地域多种果菜的精准灌溉控制，即多种果菜可以采用同一套控制系统，提高了通用性。
2.本申请为了提升需水量的预测精度，采用了一种新的激励函数-余弦激励函数，用于对整个深度神经网络的训练，极大提升了训练精度。在灌溉控制领域， 属于申请人首次提出，因此并非常规技术手段或公知常识。
3.本申请为了提升预测精度，预处理、分割、池化层、新的激励函数R l(与余弦激励函数不同的一种并列实现方案)、损失函数的使用上，都采用了新的算法，以整体上提高深度神经网络的训练的精度和速度。在灌溉控制领域，属于申请人首次提出，因此并非常规技术手段或公知常识。
4.在叶片的预处理上，针对绝大部分植株叶片都是绿色的特点，对于识别到的叶片，增强其G分量，相对抑制其R、B分量，有利于提升需求量预测判断的准确性。
5.发明人致力于果蔬精准灌溉需水量的预测研究，提出了多个并列的关联方案，而且在不会引起冲突的情况下，各个模块可以进行交叉组合，所有组合形式都构成本申请的范围之内。
发明内容
为更准确理解本发明，需先简要理解回顾下面的基本概念。
深度学习的概念由Hinton等人于2006年提出。基于深信度网(DBN)提出非监督贪心逐层训练算法，为解决深层结构相关的优化难题带来希望，随后提出多层自动编码器深层结构。此外Lecun等人提出的卷积神经网络是第一个真正多层结构学习算法，它利用空间相对关系减少参数数目以提高训练性能。
深度学习是机器学习研究中的一个新的领域，其动机在于建立、模拟人脑进行分析学习的神经网络，它模仿人脑的机制来解释数据，例如图像、声音和文本。同机器学习方法一样，深度机器学习方法也有监督学习与无监督学习之分.不同的学习框架下建立的学习模型很是不同。例如，卷积神经网络(Convolutional neural networks，简称CNNs)就是一种深度的监督学习下的机器学习模型，而深度置信网(Deep Belief Nets，简称DBNs)就是一种无监督学习下的机器学习模型。
卷积神经网络(Convolutional Neural Networks，CNN)是一类包含卷积计算且具有深度结构的前馈神经网络(Feedforward Neural Networks)，是深度学习(deep learning)的代表算法之一。
深度卷积神经网络DCNN,则是具有多个CNN层的网络结构。
深度神经网络中经常采用的激励函数如下：sigmoid函数，tanh函数，ReLU函数。
sigmoid函数，该函数是将取值为(-∞,+∞)的数映射到(0,1)之间。sigmoid函数的公式如下：
Figure PCTCN2022094818-appb-000001
sigmoid函数作为非线性激活函数，但是其并不被经常使用，它具有以下几个缺点：
(1)当z值非常大或者非常小时，sigmoid函数的导数g′(z)将接近0。这会导致权重W的梯度将接近0，使得梯度更新十分缓慢，即梯度消失。
tanh函数，tanh函数相较于sigmoid函数要常见一些，该函数是将取值为(-∞,+∞)的数映射到(-1,1)之间，其公式为：
Figure PCTCN2022094818-appb-000002
tanh函数在0附近很短一段区域内可看做线性的。由于tanh函数均值为0，因此弥补了sigmoid函数均值为0.5的缺点。
ReLU函数，ReLU函数又称为修正线性单元(Rectified Linear Unit)，是一种分段线性函数，其弥补了sigmoid函数以及tanh函数的梯度消失问题。ReLU函数的公式如下：
Figure PCTCN2022094818-appb-000003
ReLU函数的优点：
(1)在输入为正数的时候(对于大多数输入z空间来说)，不存在梯度消失问题。
(2)计算速度要快很多。ReLU函数只有线性关系，不管是前向传播还是反向传播，都比sigmod和tanh要快很多。
ReLU函数的缺点：
(1)当输入为负时，梯度为0，会产生梯度消失问题。
在能够理解上述基本概念及常规操作方式的基础上，本申请提出了一种用于日光温室果菜栽培的智能水分精准灌溉控制系统，所述系统包括：
信息采集模块，通过多种传感器进行信息采集，所采集的信息包括：土壤信息、地上环境信息、灌水设备使用过程信息、作物信息中的至少一项；
需水量预测模块，将所采集的信息输入至经训练的深度神经网络进行计算，预测出当前日光温室栽培果菜的需水量；
灌溉控制模块，根据预测出的所述果菜需水量形成控制指令，以进行定时定量的供水。
作为一种具体的实施例，所述土壤信息包括：土壤质地、土壤田间持水量、土壤温度、土壤墒情信息中的至少一项。
作为一种具体的实施例，所述地上环境信息包括：当前日光温室内部空气的温湿度信息、当前果菜冠层光照强度和光照辐射累积信息中的至少一项。
作为一种具体的实施例，所述灌水设备使用过程信息包括：灌溉管道直径信息、流速信息、灌溉时长和灌水量中的至少一项。
作为一种具体的实施例，所述作物信息包括果菜作物种类、生长阶段信息、生长状态信息。
作为一种具体的实施例，本申请所采用的传感器包括：土壤墒情传感器、光照强度传感器、空气温湿度传感器和图像传感器。
此外，本申请还提出了一种用于日光温室果菜栽培的智能水分精准灌溉控制方法，所述方法包括：
利用信息采集模块通过多种传感器进行信息采集，所采集的信息包括：土壤信息、地上环境信息、灌水设备使用过程信息、作物信息中的至少一项；
利用需水量预测模块将所采集的信息输入至经训练的深度神经网络进行计算，预测出当前日光温室栽培果菜的需水量；
利用灌溉控制模块根据预测出的所述果菜需水量形成控制指令，以进行定时定量的供水。
作为一种具体的实施例，所述土壤信息包括：土壤质地、土壤田间持水量、土壤温度、土壤墒情信息中的至少一项。
作为一种具体的实施例，所述地上环境信息包括：当前日光温室内部空气的温湿度信息、当前果菜冠层光照强度和光照辐射累积信息中的至少一项。
作为一种具体的实施例，所述灌水设备使用过程信息包括：灌溉管道直径信息、流速信息、灌溉时长和灌水量中的至少一项。
作为一种具体的实施例，所述作物信息包括果菜作物种类、生长阶段信息、 生长状态信息。
作为一种具体的实施例，本申请所采用的的传感器包括：土壤墒情传感器、光照强度传感器、空气温湿度传感器和图像传感器。
本申请还对应提出了一种计算机存储介质，所述存储介质上存储有程序代码，所述代码用于实现上述任一种所述的方法。
本申请还对应提出了一种计算机设备，所述设备包括处理器、存储器，所述存储器上存储有计算机指令，所述指令用于实现上述任一种所述的方法。
附图说明
图1表示本申请的基本实施例的结构示意图。
具体实施方式
为了使本发明的目的、技术方案和优点更加清楚，下面将结合附图对本发明作进一步地详细描述，显然，所描述的实施例仅仅是本发明一部分实施例，而不是全部的实施例。基于本发明中的实施例，本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其它实施例，都属于本发明保护的范围。
如图1所示，本申请提出了一种用于日光温室果菜栽培的智能水分精准灌溉控制系统，所述系统包括：
信息采集模块，通过多种传感器进行信息采集，所采集的信息包括：土壤信息、地上环境信息、灌水设备使用过程信息、作物信息中的至少一项；
需水量预测模块，将所采集的信息输入至经训练的深度神经网络进行计算，预测出当前日光温室栽培果菜的需水量；
灌溉控制模块，根据预测出的所述果菜需水量形成控制指令，以进行定时定量的供水。
作为一种具体的实施例，所述土壤信息包括：土壤质地、土壤田间持水量、土壤温度、土壤墒情信息中的至少一项。
作为一种具体的实施例，所述地上环境信息包括：当前日光温室内部空气的温湿度信息、当前果菜冠层光照强度和光照辐射累积信息中的至少一项。
作为一种具体的实施例，所述灌水设备使用过程信息包括：灌溉管道直径信息、流速信息、灌溉时长和灌水量中的至少一项。
作为一种具体的实施例，所述作物信息包括果菜作物种类、生长阶段信息、 生长状态信息。
作为一种具体的实施例，本申请所采用的传感器包括：土壤墒情传感器、光照强度传感器、温湿度传感器、图像传感器。
图像传感器获取果菜的图像信息，将获取的图像信息输入至果菜识别模型，获得果菜种类及生长阶段信息。
作为一种具体的实施例，需水量预测模块利用需水量预测模型获得需水量，需水量预测模型为深度卷积神经网络DCNN，所述深度卷积神经网络DCNN可以包括：信息输入层、一个或多个卷积层、一个或多个池化层、一个或多个隐含层、全连接层；所述卷积层采用的卷积核大小为3×3；所述池化层为最大池化法进行计算；所述深度卷积神经网络DCNN采用的激励函数为余弦激励函数，记为f(),其中
Figure PCTCN2022094818-appb-000004
其中，θ yi表示为样本i与其对应标签y i的向量夹角；所述N表示训练样本个数；所述w yi表示样本i在其标签y i处的权重；
所述信息输入层接收来自一个或多个传感器的信息；
果蔬的需水量预测模型接收来自果菜识别模型的输出结果，包括：果菜种类、果菜生长阶段、果菜生长状态中的至少一项信息。果菜识别模型采用了VGG-16网络模型。利用训练集进行训练，训练集包含多种不同品种果菜，包括：黄瓜、番茄、辣椒、茄子等100多种果蔬图像，共25783张图像，其中75％作为训练集，25％作为测试集。为了加速训练过程，还使用了AdamOptimizer作为优化器，分别采用神经元为64×32、32×16的卷积层，使用3×3卷积核，步长为2的最大池化层，64个神经元的全连接层和softmax回归层进行实现。所述优化器包括：GradientDescentOptimizer、AdagradOptimizer、AdagradDAOptimizer、MomentumOptimizer、RMSPropOptimizer中的至少一个。
进一步，可选的，对所述获取的果菜的图像信息进行预处理，所述预处理还包括：图像信息的筛选，剔除不满足清晰度要求或未能包含至少一片完整叶片的图像；所述图像裁剪包括：将获得的图像信息进行裁剪，获得叶片信息。
进一步，可选的，对于叶片为绿色的果蔬，所述预处理还包括：将RGB空 间中的R与B分量进行色彩抑制，将G通道分量进行增强。
进一步，可选的，对G分量进行增强，首先对分割后的叶片进行R、G、B分量的分离；生成G分量的调节系数：
Figure PCTCN2022094818-appb-000005
然后
利用调节系数进行调整：
Figure PCTCN2022094818-appb-000006
利用调整后的G F逆向合成叶片图像；
其中各分量的分解及合成属于本领域的现有技术，但是上述调节方式是本发明人的首创，经过上述调整，能够充分利用叶片的绿色通道信息，使得后续的预测经过更加准确。
进一步，可选的，所述需水量预测模型具体包括多区域卷积神经网络模型，所述多区域卷积神经网络模型包括：卷积网络层，用于生成原始叶片的映射特征；多区域置信网络模型，包括多个区域的置信网络模型，用于对所述果菜的当前状态生成不同需水量的多个不同置信度值，对多个区域的不同置信度值进行拟合，确定出在不同区域中置信度值都相对较大的置信度值，将其对应的需水量确定为果菜的需水量。
进一步，可选的，所述多区域置信网络模型包括多区域池化层和完全连接层，多区域池化层包括多个区域的池化层，池化层个数为1，池化层还可替换成完全连接层；所述池化层用于生成置信度。
进一步，可选的，所述图像分割采用了改进的分水岭分割方式进行分割：
Figure PCTCN2022094818-appb-000007
其中Gradient(x,y)表示像素点(x,y)的原始梯度值；
Figure PCTCN2022094818-appb-000008
Figure PCTCN2022094818-appb-000009
分别表示在窗口D区域内的梯度均值、梯度最小值、梯度最大值；Gra表示修正梯度值；
S＝watershed(Gra)
进一步，可选的，所述需水量预测模型包括卷积网络层及多区域置信网络，所述卷积网络层，用于生成分割出的各叶片特征；多区域置信网络为包括多个叶片区域的置信网络模型，用于对多个不同叶片缺水程度产生置信度值，当至少两 个叶片区域的置信度值均满足预设的第一阈值等级范围时，以此时的缺水量作为果菜的需水量。
进一步，可选的，作为另一实施例，所述需水量预测模型包括卷积网络层及多区域置信网络，所述卷积网络层能够接收所拍摄图像的分割后叶片图像，并生成不同尺度的卷积多特征映射；所述多区域置信网络模型包括多区域池化层和完全连接层；其中，所述多区域池化层包括多个区域池化层；所述多个区域池化层用于生成果菜图像需水量的置信度值，当至少两个叶片区域的置信度值的置信度值均满足预设的第一阈值等级范围时，则确定此时的需水量作为果菜的需水量；所述多区域池化层设置为最大池化层或平均池化层；所述完全连接层用于对识别到的叶片缺水程度进行分类。
可选的，作为另一个实施例：所述需水量预测模型为深度卷积神经网络，具体包括：输入层、嵌入层、池化层、全连接层；所述输入层用于接收拍摄图像的分割后叶片图像；所述嵌入层采用的卷积核大小为5×5；激励函数记为R l()；经过全连接层处理后进一步得到需水量预测推荐结果；
所述池化层的池化方法如下：
Figure PCTCN2022094818-appb-000010
其中，x e表示当前层的输出，u e表示激励函数R l的输入，
R l()表示激励函数，w e表示当前层的权重，φ表示损失函数，x e-1表示上一层的输出；
激励函数R l为：
Figure PCTCN2022094818-appb-000011
所述损失函数φ如下：
Figure PCTCN2022094818-appb-000012
Figure PCTCN2022094818-appb-000013
N表示正样本数据集的大小，i取值1～N，y i表示正样本x i对应的标签值；W yi表示正样本特征向量x i在其标签y i处的权重，s为深度卷积神经网络的推荐 参数；b j表示样本x i在其标签y i处的偏差。
此外，本申请还提出了一种用于日光温室果菜栽培的智能水分精准灌溉控制方法，所述方法包括：
利用信息采集模块通过多种传感器进行信息采集，所采集的信息包括：土壤信息、地上环境信息、灌水设备使用过程信息、作物信息中的至少一项；
利用需水量预测模块将所采集的信息输入至经训练的深度神经网络进行计算，预测出当前日光温室栽培果菜的需水量；
利用灌溉控制模块根据预测出的所述果菜需水量形成控制指令，以进行定时定量的供水。
作为一种具体的实施例，所述土壤信息包括：土壤质地、土壤田间持水量、土壤温度、土壤墒情信息中的至少一项。
作为一种具体的实施例，所述地上环境信息包括：当前日光温室内部空气的温湿度信息、当前果菜冠层光照强度和光照辐射累积信息中的至少一项。
作为一种具体的实施例，所述灌水设备使用过程信息包括：灌溉管道直径信息、流速信息、灌溉时长和灌水量中的至少一项。
作为一种具体的实施例，所述作物信息包括果菜作物种类、生长阶段信息、生长状态信息。
作为一种具体的实施例，本申请所采用的传感器包括：土壤墒情传感器、光照强度传感器、温湿度传感器、图像传感器。
图像传感器获取果菜的图像信息，将获取的图像信息输入至果菜识别模型，获得果菜种类及生长阶段信息。
作为一种具体的实施例，需水量预测模块利用需水量预测模型获得需水量，需水量预测模型为深度卷积神经网络DCNN，所述深度卷积神经网络DCNN包括：信息输入层、一个或多个卷积层、一个或多个池化层、一个或多个隐含层、全连接层；所述卷积层采用的卷积核大小为3×3；所述池化层为最大池化法进行计算；所述深度卷积神经网络DCNN采用的激励函数为余弦激励函数，记为f(),其中
Figure PCTCN2022094818-appb-000014
其中，θ yi表示为样本i与其对应标签y i的向量夹角；所述N表示训练样本个 数；所述w yi表示样本i在其标签y i处的权重。
所述信息输入层接收来自一个或多个传感器的信息；
果菜的需水量预测模型接收来自果菜识别模型的输出结果，包括：果菜种类、果菜生长阶段、果菜生长状态中的至少一项信息。果菜识别模型采用了VGG-16网络模型。利用训练集进行训练，训练集包含多种不同品种果菜，包括：茄子、番茄、青瓜、白菜、土豆等100多种果蔬图像，共25783张图像，其中75％作为训练集，25％作为测试集。为了加速训练过程，还使用了优化器进行优化，分别采用神经元为64×32、32×16的卷积层，使用3×3卷积核，步长为2的最大池化层，64个神经元的全连接层和softmax回归层进行实现。所述优化器包括：GradientDescentOptimizer、AdagradOptimizer、AdagradDAOptimizer、MomentumOptimizer、RMSPropOptimizer中的至少一个。
进一步，可选的，对所述获取的果菜的图像信息进行预处理，所述预处理还包括：图像信息的筛选，剔除不满足清晰度要求或未能包含至少一片完整叶片的图像；所述图像裁剪包括：将获得的图像信息进行裁剪，获得叶片信息。
进一步，可选的，对于叶片为绿色的果蔬，所述预处理还包括：将RGB空间中的R与B分量进行色彩抑制，将G通道分量进行增强。
进一步，可选的，对G分量进行增强，首先对分割后的叶片进行R、G、B分量的分离；生成G分量的调节系数：
Figure PCTCN2022094818-appb-000015
然后利用调节系数进行调整：
Figure PCTCN2022094818-appb-000016
利用调整后的G F逆向合成叶片图像；
其中各分量的分解及合成属于本领域的现有技术，但是上述调节方式是本发明人的首创，经过上述调整，能够充分利用叶片的绿色通道信息，使得后续的预测经过更加准确。
进一步，可选的，作为一具体的实施例，所述需水量预测模型具体包括多区域卷积神经网络模型，所述多区域卷积神经网络模型包括：卷积网络层，用于生成原始叶片的映射特征；多区域置信网络模型，包括多个区域的置信网络模型，用于对所述果菜的当前状态生成不同需水量的多个不同置信度值，对多个区域的不同置信度值进行拟合，确定出在不同区域中置信度值都相对较大的置信度值， 将其对应的需水量确定为果菜的需水量。
进一步，可选的，所述多区域置信网络模型包括多区域池化层和完全连接层，多区域池化层包括多个区域的池化层，池化层个数为1，池化层还可替换成完全连接层；所述池化层用于生成置信度。
进一步，可选的，所述图像分割采用了改进的分水岭分割方式进行分割：
Figure PCTCN2022094818-appb-000017
其中Gradient(x,y)表示像素点(x,y)的原始梯度值；
Figure PCTCN2022094818-appb-000018
Figure PCTCN2022094818-appb-000019
分别表示在窗口D区域内的梯度均值、梯度最小值、梯度最大值；Gra表示修正梯度值；
S＝watershed(Gra)
进一步，可选的，所述需水量预测模型包括卷积网络层及多区域置信网络，所述卷积网络层，用于生成分割出的各叶片特征；多区域置信网络为包括多个叶片区域的置信网络模型，用于对多个不同叶片缺水程度产生置信度值，当至少两个叶片区域的置信度值均满足预设的第一阈值等级范围时，以此时的缺水量作为果菜的需水量。
进一步，可选的，作为另一实施例，所述需水量预测模型包括卷积网络层及多区域置信网络，所述卷积网络层能够接收所拍摄图像的分割后叶片图像，并生成不同尺度的卷积多特征映射；所述多区域置信网络模型包括多区域池化层和完全连接层；其中，所述多区域池化层包括多个区域池化层；所述多个区域池化层用于生成果菜图像需水量的置信度值，当至少两个叶片区域的置信度值的置信度值均满足预设的第一阈值等级范围时，则确定此时的需水量作为果菜的需水量；所述多区域池化层设置为最大池化层或平均池化层；所述完全连接层用于对识别到的叶片缺水程度进行分类。
可选的，作为另一个实施例，所述需水量预测模型为深度卷积神经网络，具体包括：输入层、嵌入层、池化层、全连接层；所述输入层用于接收拍摄图像的分割后叶片图像；所述嵌入层采用的卷积核大小为5×5；激励函数记为R l()；经过全连接层处理后进一步得到需水量预测推荐结果；
所述池化层的池化方法如下：
Figure PCTCN2022094818-appb-000020
其中，x e表示当前层的输出，u e表示激励函数R l的输入，
R l()表示激励函数，w e表示当前层的权重，φ表示损失函数，x e-1表示上一层的输出；
激励函数R l为：
Figure PCTCN2022094818-appb-000021
所述损失函数φ如下：
Figure PCTCN2022094818-appb-000022
Figure PCTCN2022094818-appb-000023
N表示正样本数据集的大小，i取值1～N，y i表示正样本x i对应的标签值；W yi表示正样本特征向量x i在其标签y i处的权重，s为深度卷积神经网络的推荐参数；b j表示样本x i在其标签y i处的偏差。
在本说明书的描述中，参考术语“一个实施例”、“示例”、“具体示例”等的描述意指结合该实施例或示例描述的具体特征、结构、材料或者特点包含于本发明的至少一个实施例或示例中。在本说明书中，对上述术语的示意性表述不一定指的是相同的实施例或示例。
本申请还提出一种计算机可读介质，上面包含可实现上述系统的程序代码，所包含的程序代码可以用任何适当的介质传输，包括但不限于无线、电线、光缆、RF等等，或者上述的任意合适的组合。计算机可读介质可以是计算机可读信号介质或者计算机可读存储介质。计算机可读存储介质例如可以是电、磁、光、电磁、红外线、或半导体的系统、装置或器件，或者任意以上的组合。计算机可读存储介质的更具体的例子(非穷举的列表)包括：具有一个或多个导线的电连接、便携式计算机磁盘、硬盘、随机存取存储器(RAM)、只读存储器(ROM)、可擦式可编程只读存储器(EPROM或闪存)、光纤、便携式紧凑磁盘只读存储器(CD-ROM)、光存储器件、磁存储器件、或者上述的任意合适的组合。在本文件 中，计算机可读存储介质可以是任何包含或存储程序的有形介质，该程序可以被指令执行系统、装置或者器件使用或者与其结合使用。可以以一种或多种程序设计语言或其组合来编写用于执行本发明操作的计算机程序代码，所述程序设计语言包括面向对象的程序设计语言—诸如Java、Smalltalk、C++，还包括常规的过程式程序设计语言—诸如“C”语言或类似的程序设计语言。程序代码可以完全地在用户计算机上执行、部分地在用户计算机上执行、作为一个独立的软件包执行、部分在用户计算机上部分在远程计算机上执行、或者完全在远程计算机或服务器上执行。在涉及远程计算机的情形中，远程计算机可以通过任意种类的网络——包括局域网(LAN)或广域网(WAN)连接到用户计算机，或者，可以连接到外部计算机(例如利用因特网服务提供商来通过因特网连接)。上述以软件功能单元的形式实现的集成的单元，可以存储在一个计算机可读取存储介质中。上述软件功能单元存储在一个存储介质中，包括若干指令用以使得一台计算机设备(可以是个人计算机，服务器，或者网络设备等)或处理器(processor)执行本发明各个实施例所述方法的部分步骤。而前述的存储介质包括：U盘、移动硬盘、只读存储器(Read-Only Memory，ROM)、随机存取存储器(Random Access Memory，RAM)、磁碟或者光盘等各种可以存储程序代码的介质。
以上所述仅为本发明的优选实施例，并非因此限制本发明的专利范围，凡是利用本发明说明书及附图内容所作的等效结构或等效流程变换，或直或间接运用在其他相关的技术领域，均同理包括在本发明的专利保护范围内。以上公开的本发明优选实施例只是用于帮助阐述本发明。优选实施例并没有详尽叙述所有的细节，也不限制该发明仅为所述的具体实施方式。显然，根据本说明书的内容，可作很多的修改和变化。本说明书选取并具体描述这些实施例，是为了更好地解释本发明的原理和实际应用，从而使所属技术领域技术人员能很好地理解和利用本发明。本发明仅受权利要求书及其全部范围和等效物的限制。

Claims (10)

  1. 一种用于日光温室果菜栽培的智能水分精准灌溉控制系统，所述系统包括：
    信息采集模块，通过多种传感器进行信息采集，所采集的信息包括：土壤信息、地上环境信息、灌水设备使用过程信息、作物信息中的至少一项；
    需水量预测模块，将所采集的信息输入至经训练的深度神经网络进行计算，预测出当前日光温室栽培果菜的需水量；
    灌溉控制模块，根据预测出的所述果菜的需水量形成控制指令，以进行定时定量的供水；
    需水量预测模块利用需水量预测模型获得需水量，需水量预测模型为深度卷积神经网络DCNN，所述深度卷积神经网络DCNN可以包括：信息输入层、一个或多个卷积层、一个或多个池化层、一个或多个隐含层、全连接层；所述卷积层采用的卷积核大小为3*3；所述池化层为最大池化法进行计算；所述深度卷积神经网络DCNN采用的激励函数为余弦激励函数，记为f(),其中
    Figure PCTCN2022094818-appb-100001
    其中，θ yi表示为样本i与其对应标签y i的向量夹角；所述N表示训练样本个数；所述w yi表示样本i在其标签y i处的权重；
    对于叶片为绿色的果蔬，预处理还包括：将RGB空间中的R与B分量进行色彩抑制，将G通道分量进行增强；
    首先对分割后的叶片进行R、G、B分量的分离；生成G分量的调节系数：
    Figure PCTCN2022094818-appb-100002
    然后利用调节系数进行调整：
    Figure PCTCN2022094818-appb-100003
    利用调整后的G F逆向合成叶片图像。
  2. 根据权利要求1所述的控制系统，所述土壤信息包括：土壤质地、土壤田间持水量、土壤温度、土壤墒情信息中的至少一项。
  3. 根据权利要求1所述的控制系统，所述地上环境信息包括：当前日光温室内部空气的温湿度信息、当前果菜冠层光照强度和光照辐射累积信息中的至少一项。
  4. 根据权利要求1所述的控制系统，所述灌水设备使用过程信息包括：灌溉管道直径信息、流速信息、灌溉时长和灌水量中的至少一项。
  5. 根据权利要求1所述的控制系统，所述作物信息包括果菜作物种类、阶段信息、生长状态信息。
  6. 一种用于日光温室果菜栽培的智能水分精准灌溉控制方法，所述方法包括：
    利用信息采集模块通过多种传感器进行信息采集，所采集的信息包括：土壤信息、地上环境 信息、灌水设备使用过程信息、作物信息中的至少一项；
    利用需水量预测模块将所采集的信息输入至经训练的深度神经网络进行计算，预测出当前日光温室栽培果菜的需水量；
    利用灌溉控制模块根据预测出的所述果菜需水量形成控制指令，以进行定时定量的供水所述需水量预测模型具体包括多区域卷积神经网络模型，所述多区域卷积神经网络模型包括：卷积网络层，用于生成原始叶片的映射特征；多区域置信网络模型，包括多个区域的置信网络模型，用于对所述果菜的当前状态生成不同需水量的多个不同置信度值，对多个区域的不同置信度值进行拟合，确定出在不同区域中置信度值都相对较大的置信度值，将其对应的需水量确定为果菜的需水量；
    所述多区域置信网络模型包括多区域池化层和完全连接层，多区域池化层包括多个区域的池化层，池化层个数为1；所述池化层用于生成置信度；
    图像分割采用了改进的分水岭分割方式进行分割：
    Figure PCTCN2022094818-appb-100004
    其中Gradient(x,y)表示像素点(x,y)的原始梯度值；
    Figure PCTCN2022094818-appb-100005
    Figure PCTCN2022094818-appb-100006
    分别表示在窗口D区域内的梯度均值、梯度最小值、梯度最大值；Gra表示修正梯度值；
    S＝watershed(Gra)，S表示最终的分割结果。
  7. 根据权利要求6所述的控制方法，所述土壤信息包括：土壤质地、土壤田间持水量、土壤温度、土壤墒情信息中的至少一项。
  8. 根据权利要求6所述的控制方法，所述地上环境信息包括：当前日光温室内部空气的温湿度信息、当前果菜冠层光照强度和光照辐射累积信息中的至少一项。
  9. 根据权利要求6所述的控制方法，所述灌水设备使用过程信息包括：灌溉管道直径信息、流速信息、灌溉时长和灌水量中的至少一项。
  10. 根据权利要求6所述的控制方法，所述作物信息包括果菜作物种类、生长阶段信息、生长状态信息。
PCT/CN2022/094818 2021-06-01 2022-05-25 一种用于日光温室果菜栽培的智能水分精准灌溉控制系统及方法 WO2022253057A1 (zh)

Applications Claiming Priority (4)

Application Number Priority Date Filing Date Title
CN202110607031.3 2021-06-01
CN202110607031.3A CN113141940A (zh) 2021-06-01 2021-06-01 一种用于日光温室果菜栽培的智能水分精准灌溉控制系统及方法
CN202111041109.6A CN113557890B (zh) 2021-06-01 2021-09-06 一种用于日光温室果菜栽培的智能水分精准灌溉控制系统及方法
CN202111041109.6 2021-09-06

Publications (1)

Publication Number Publication Date
WO2022253057A1 true WO2022253057A1 (zh) 2022-12-08

Family

ID=76875409

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2022/094818 WO2022253057A1 (zh) 2021-06-01 2022-05-25 一种用于日光温室果菜栽培的智能水分精准灌溉控制系统及方法

Country Status (2)

Country Link
CN (2) CN113141940A (zh)
WO (1) WO2022253057A1 (zh)

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116188196A (zh) * 2023-03-13 2023-05-30 华厦筑工(广州)技术有限公司 一种智慧园区设备可视化运营系统
CN116210565A (zh) * 2023-03-29 2023-06-06 上海华维可控农业科技集团股份有限公司 一种用于智慧灌溉下的回液处理系统及方法
CN117743975A (zh) * 2024-02-21 2024-03-22 君研生物科技(山西)有限公司 山坡耕地土壤环境改良方法
CN117882546A (zh) * 2024-03-13 2024-04-16 山西诚鼎伟业科技有限责任公司 一种面向农业作业机器人的智能化种植方法
CN117981663A (zh) * 2024-04-07 2024-05-07 黑龙江智云互联农业科技有限公司 水稻田自动化节水精准灌溉系统
CN118036830A (zh) * 2024-04-11 2024-05-14 成都信息工程大学 多熟栽培模式下花魔芋产量预测方法

Families Citing this family (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113141940A (zh) * 2021-06-01 2021-07-23 中国农业科学院蔬菜花卉研究所 一种用于日光温室果菜栽培的智能水分精准灌溉控制系统及方法
CN114451257B (zh) * 2021-12-23 2023-04-07 珠海格力电器股份有限公司 基于神经网络的灌溉方法、装置、存储介质及电子设备
CN114723190B (zh) * 2022-06-09 2022-08-23 苏州农业职业技术学院 一种生产栽培中的月季花期调控方法及系统
CN114878788A (zh) * 2022-06-14 2022-08-09 古田县恒春农业开发有限公司 一种花卉种植智能湿度检测装置
CN115530054A (zh) * 2022-10-12 2022-12-30 河北省科学院应用数学研究所 灌溉控制方法、装置、电子设备及存储介质
CN115777495A (zh) * 2022-12-05 2023-03-14 贵州省水利科学研究院(贵州省灌溉试验中心站) 基于参考作物需水量的烤烟栽培滴灌系统、设备及终端
CN115989763B (zh) * 2023-03-23 2023-06-27 北京市农林科学院智能装备技术研究中心 温室灌溉控制方法、装置、系统及设备
CN116058195B (zh) * 2023-04-06 2023-07-04 中国农业大学 一种叶菜生长环境光照调控方法、系统及装置
CN116609672B (zh) * 2023-05-16 2024-05-07 国网江苏省电力有限公司淮安供电分公司 一种基于改进的bwoa-fnn算法的储能电池soc估计方法

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109845625A (zh) * 2018-12-12 2019-06-07 珠江水利委员会珠江水利科学研究院 一种基于神经网络的多维参量农作物智能灌溉控制方法
WO2019118460A1 (en) * 2017-12-11 2019-06-20 The Texas A&M University System Irrigation system control with predictive water balance capabilities
CN109978234A (zh) * 2019-03-01 2019-07-05 北京农业信息技术研究中心 墒情预测、灌溉决策方法和装置
CN112042353A (zh) * 2020-09-03 2020-12-08 山东锋士信息技术有限公司 一种适用于日光温室的水肥精准决策方法及系统
CN113557890A (zh) * 2021-06-01 2021-10-29 中国农业科学院蔬菜花卉研究所 一种用于日光温室果菜栽培的智能水分精准灌溉控制系统及方法

Family Cites Families (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
AU2015404215B2 (en) * 2015-08-06 2019-01-03 Accenture Global Services Limited Vegetation management for power line corridor monitoring using computer vision
CN204884141U (zh) * 2015-08-27 2015-12-16 钦州学院 基于物联网的农田智能灌溉控制系统
CN105046924A (zh) * 2015-08-27 2015-11-11 钦州学院 基于物联网的农田智能灌溉控制系统及其运行方法
KR101876397B1 (ko) * 2016-11-23 2018-07-09 세종대학교산학협력단 작물의 병해충 진단 장치 및 방법
CN106707757A (zh) * 2017-01-23 2017-05-24 中国农业大学 一种灌水时间动态调控方法及其系统
CN108617480A (zh) * 2018-05-11 2018-10-09 武汉理工大学 一种园林智能浇灌系统
CN110050673A (zh) * 2019-04-30 2019-07-26 黄河水利委员会黄河水利科学研究院 一种智能灌溉管理系统
CN111626518A (zh) * 2020-05-29 2020-09-04 上海交通大学 基于深度学习神经网络的城市天级需水量在线预测方法
CN112733710A (zh) * 2021-01-08 2021-04-30 徐州展皎信息技术服务有限公司 用于灌溉装置的灌溉水压控制的神经网络的训练方法

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2019118460A1 (en) * 2017-12-11 2019-06-20 The Texas A&M University System Irrigation system control with predictive water balance capabilities
CN109845625A (zh) * 2018-12-12 2019-06-07 珠江水利委员会珠江水利科学研究院 一种基于神经网络的多维参量农作物智能灌溉控制方法
CN109978234A (zh) * 2019-03-01 2019-07-05 北京农业信息技术研究中心 墒情预测、灌溉决策方法和装置
CN112042353A (zh) * 2020-09-03 2020-12-08 山东锋士信息技术有限公司 一种适用于日光温室的水肥精准决策方法及系统
CN113557890A (zh) * 2021-06-01 2021-10-29 中国农业科学院蔬菜花卉研究所 一种用于日光温室果菜栽培的智能水分精准灌溉控制系统及方法

Cited By (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116188196A (zh) * 2023-03-13 2023-05-30 华厦筑工(广州)技术有限公司 一种智慧园区设备可视化运营系统
CN116188196B (zh) * 2023-03-13 2024-01-02 华厦筑工(广州)技术有限公司 一种智慧园区设备可视化运营系统
CN116210565A (zh) * 2023-03-29 2023-06-06 上海华维可控农业科技集团股份有限公司 一种用于智慧灌溉下的回液处理系统及方法
CN116210565B (zh) * 2023-03-29 2023-11-14 上海华维可控农业科技集团股份有限公司 一种用于智慧灌溉下的回液处理系统及方法
CN117743975A (zh) * 2024-02-21 2024-03-22 君研生物科技(山西)有限公司 山坡耕地土壤环境改良方法
CN117882546A (zh) * 2024-03-13 2024-04-16 山西诚鼎伟业科技有限责任公司 一种面向农业作业机器人的智能化种植方法
CN117882546B (zh) * 2024-03-13 2024-05-24 山西诚鼎伟业科技有限责任公司 一种面向农业作业机器人的智能化种植方法
CN117981663A (zh) * 2024-04-07 2024-05-07 黑龙江智云互联农业科技有限公司 水稻田自动化节水精准灌溉系统
CN118036830A (zh) * 2024-04-11 2024-05-14 成都信息工程大学 多熟栽培模式下花魔芋产量预测方法

Also Published As

Publication number Publication date
CN113557890B (zh) 2022-01-28
CN113141940A (zh) 2021-07-23
CN113557890A (zh) 2021-10-29

Similar Documents

Publication Publication Date Title
WO2022253057A1 (zh) 一种用于日光温室果菜栽培的智能水分精准灌溉控制系统及方法
WO2023024615A1 (zh) 一种基于植株萎蔫程度进行精准灌溉的系统及方法
US20230409910A1 (en) Ai-powered autonomous plant-growth optimization system that automatically adjusts input variables to yield desired harvest traits
Khairunniza-Bejo et al. Application of artificial neural network in predicting crop yield: A review
US20220075344A1 (en) A method of finding a target environment suitable for growth of a plant variety
Quiroz et al. Image recognition of Legacy blueberries in a Chilean smart farm through deep learning
CN117036088A (zh) 一种ai识别绿化植物生长态势的数据采集分析方法
CN109978234A (zh) 墒情预测、灌溉决策方法和装置
CN102789579A (zh) 基于计算机视觉技术的温室作物水肥胁迫状态识别方法
CN110119169A (zh) 一种基于最小向量机的番茄温室温度智能预警系统
CN114027052A (zh) 一种植物生殖发育的光照调控系统
Kurtulmuş et al. Deep learning for proximal soil sensor development towards smart irrigation
Veeragandham et al. A review on the role of machine learning in agriculture
Ting et al. Universality of an improved photosynthesis prediction model based on PSO-SVM at all growth stages of tomato
Barriga et al. Crop-water assessment in Citrus (Citrus sinensis L.) based on continuous measurements of leaf-turgor pressure using machine learning and IoT
Lu et al. Image classification and identification for rice leaf diseases based on improved WOACW_SimpleNet
Lin et al. Data-driven modeling for crop growth in plant factories
Gao et al. How can agricultural water production be promoted? A review on machine learning for irrigation
Jain et al. Performance prediction for crop irrigation using different machine learning approaches
Ye et al. Estimation of rice seedling growth traits with an end-to-end multi-objective deep learning framework
CN116957141A (zh) 一种基于Transformer模型的智能灌溉预测方法及系统
Macabiog et al. Soil moisture and rain prediction based irrigation controller for the strawberry farm of La Trinidad, Benguet
CN115510991A (zh) 一种基于1dcnn-lstm多源信息融合模型的日光温室蚜虫预警方法
CN110210376A (zh) 一种基于深度学习的高分遥感影像西北地区农膜提取方法
Bobade et al. Design of Smart Irrigation System Based on MLA

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 22815103

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 22815103

Country of ref document: EP

Kind code of ref document: A1