A method, device and Internet of Things system for controlling sprinkler irrigation

Abstract

This invention discloses a sprinkler irrigation control method, device, and Internet of Things (IoT) system. The sprinkler irrigation control IoT system includes moisture sensing and control terminals 1-n, a sprinkler irrigation transmission terminal, a sprinkler irrigation cloud server, and a sprinkler irrigation application terminal. The sprinkler irrigation application terminal includes a sprinkler irrigation monitoring terminal and a sprinkler irrigation interactive APP. Each moisture sensing and control terminal wirelessly interacts with the sprinkler irrigation cloud server through the sprinkler irrigation transmission terminal. The sprinkler irrigation monitoring terminal and the sprinkler irrigation interactive APP are used to facilitate remote monitoring of the sprinkler irrigation system by users, enabling precise adjustment of the sprinkler irrigation device. This invention has the advantage of high moisture regulation accuracy.

Description

Technical Field

[0001] This invention belongs to the field of sprinkler irrigation detection and control technology, specifically relating to a sprinkler irrigation control method, device and Internet of Things system. Background Technology

[0002] Sprinkler irrigation is used in many applications, such as industrial processing, environmental purification, agricultural crops, and flower and bonsai cultivation.

[0003] There are two main existing methods for controlling sprinkler irrigation:

[0004] (1) Timed sprinkler irrigation is simple to control and low in cost, but it cannot replenish water reasonably according to demand;

[0005] (2) A moisture sensor is used to monitor moisture in real time and an automatic sprinkler irrigation is realized based on a PID algorithm. The moisture sensor detection value is directly used as the input of the PID algorithm. However, the moisture sensor detection has problems with poor stability and accuracy, which affects the accuracy of moisture regulation during sprinkler irrigation. In addition, the PID algorithm is based on well-tuned control parameters and is suitable for relatively stable working conditions. When the environment changes, it cannot adaptively adjust the control parameters, resulting in poor flexibility in moisture regulation and affecting the accuracy of moisture regulation.

[0006] On the other hand, there are mutual influences between adjacent sprinkler irrigation areas. For example, after water seeps into the soil, it will penetrate and spread between sprinkler irrigation areas, and the existing sprinkler irrigation control methods do not take this factor into account.

[0007] In summary, existing sprinkler irrigation control methods suffer from low precision in water regulation. Summary of the Invention

[0008] Purpose of the invention: The purpose of this invention is to provide a spray irrigation control method, device, and Internet of Things system with high accuracy in water regulation.

[0009] Technical solution: The sprinkler irrigation control method of the present invention includes a sprinkler irrigation area comprising adjacent sprinkler irrigation area 1 and sprinkler irrigation area 2, and the sprinkler irrigation control method comprising:

[0010] Acquire the outputs of multiple moisture sensors in sprinkler irrigation area 1 and multiple moisture sensors in sprinkler irrigation area 2;

[0011] Constructing a sprinkler irrigation control model:

[0012] The sprinkler irrigation control model includes NARX neural network models 1-2, DRNN neural network models 1-2, sprinkler irrigation controllers 1-2, moisture detection modules 1-2, fuzzy RBF neural network model-DRNN neural network model 1, and fuzzy RBF neural network model-NARX neural network model; wherein, moisture detection module 1 is used to obtain the moisture detection value of sprinkler irrigation area 1 based on the output of multiple moisture sensors in sprinkler irrigation area 1, and the moisture detection value of sprinkler irrigation area 1 is used as the input of sprinkler irrigation controller 1; moisture detection module 2 is used to obtain the moisture detection value of sprinkler irrigation area 2 based on the output of multiple moisture sensors in sprinkler irrigation area 2, and the moisture detection value of sprinkler irrigation area 2 is used as the input of sprinkler irrigation controller 2;

[0013] The target moisture value of the sprinkler irrigation area 1 is used as the input of the NARX neural network model 1 and the moisture control value of the sprinkler irrigation controller 1, respectively. The sum of the output of the sprinkler irrigation controller 1 and the output of the NARX neural network model 1 is used as the corresponding input of the DRNN neural network model 1 and the corresponding input of the fuzzy RBF neural network model-DRNN neural network model 1, respectively. The output of the DRNN neural network model 1 is used as the input of the sprinkler irrigation device 1. The sprinkler irrigation device 1 is used to adjust the moisture of the sprinkler irrigation area 1.

[0014] The target moisture value of the sprinkler irrigation area 2 is used as the input of NARX neural network model 2 and the corresponding input of fuzzy RBF neural network model-DRNN neural network model 1, respectively. The output of fuzzy RBF neural network model-DRNN neural network model 1 is used as the moisture control value of sprinkler irrigation controller 2. The sum of the output of sprinkler irrigation controller 2 and the output of NARX neural network model 2 is used as the corresponding input of DRNN neural network model 2. The output of DRNN neural network model 2 is used as the input of sprinkler irrigation device 2. Sprinkler irrigation device 2 is used to adjust the moisture of sprinkler irrigation area 2.

[0015] The outputs of DRNN neural network model 1 and DRNN neural network model 2 are used as the corresponding inputs of the fuzzy RBF neural network model-NARX neural network model, respectively. The outputs of moisture detection module 1 and moisture detection module 2 are used as the corresponding inputs of the fuzzy RBF neural network model-NARX neural network model, respectively. The moisture disturbance amount of the sprinkler irrigation area 1 output by the fuzzy RBF neural network model-NARX neural network model is used as the corresponding input of DRNN neural network model 1, and the moisture disturbance amount of the sprinkler irrigation area 2 output is used as the corresponding input of DRNN neural network model 2.

[0016] Furthermore, the sprinkler irrigation area also includes a sprinkler irrigation area 3 adjacent to the sprinkler irrigation area 2, and the sprinkler irrigation control method further includes:

[0017] Acquire the outputs of three moisture sensors in the sprinkler irrigation area;

[0018] The sprinkler irrigation control model also includes NARX neural network model 3, DRNN neural network model 3, sprinkler irrigation controller 3, moisture detection module 3 and fuzzy RBF neural network model-DRNN neural network model 2; the moisture detection module 3 is used to obtain the moisture detection value of the sprinkler irrigation area 3 based on the output of multiple moisture sensors in the sprinkler irrigation area 3, and the moisture detection value of the sprinkler irrigation area 3 is used as the input of the sprinkler irrigation controller 3.

[0019] The sum of the outputs of the sprinkler irrigation controller 2 and the NARX neural network model 2 is also used as the corresponding input of the fuzzy RBF neural network model - DRNN neural network model 2;

[0020] The target moisture value of the sprinkler irrigation area 3 is used as the input of NARX neural network model 3 and the corresponding input of fuzzy RBF neural network model-DRNN neural network model 2, respectively. The output of fuzzy RBF neural network model-DRNN neural network model 2 is used as the moisture control value of sprinkler irrigation controller 3. The sum of the output of sprinkler irrigation controller 3 and the output of NARX neural network model 3 is used as the corresponding input of DRNN neural network model 3. The output of DRNN neural network model 3 is used as the input of sprinkler irrigation device 3. Sprinkler irrigation device 3 is used to adjust the moisture of sprinkler irrigation area 3.

[0021] The output of DRNN neural network model 3 serves as the corresponding input of the fuzzy RBF neural network model-NARX neural network model, the output of moisture detection module 3 serves as the corresponding input of the fuzzy RBF neural network model-NARX neural network model, and the moisture disturbance amount of the sprinkler irrigation area 3 output by the fuzzy RBF neural network model-NARX neural network model serves as the corresponding input of DRNN neural network model 3.

[0022] The above provides technical solutions for two and three sprinkler irrigation areas. It is understood that, under the technical concept of this invention, sprinkler irrigation control models applicable to more adjacent sprinkler irrigation areas can also be designed.

[0023] Furthermore, the moisture detection module includes DRNN neural network models 1-3, BiGRU neural network models 1-3, and fuzzy RBF neural network model-NARX neural network model;

[0024] The outputs of multiple moisture sensors serve as the corresponding inputs to DRNN neural network model 1, DRNN neural network model 2, and DRNN neural network model 3, respectively. The output of DRNN neural network model 1 serves as the corresponding input to BiGRU neural network model 1 and fuzzy RBF neural network model-NARX neural network model, respectively. The output of DRNN neural network model 2 serves as the corresponding input to BiGRU neural network model 2 and fuzzy RBF neural network model-NARX neural network model, respectively. The outputs of BiGRU neural network model 1, BiGRU neural network model 2, and BiGRU neural network model 3 serve as the corresponding input to fuzzy RBF neural network model-NARX neural network model, respectively. The output of fuzzy RBF neural network model-NARX neural network model is used as the moisture detection value of the sprinkler irrigation area.

[0025] Furthermore, the sprinkler irrigation controller includes a fuzzy RBF neural network model-PID controller, a NARX neural network model-PID controller, a DRNN neural network model-PID controller, a NARX neural network model, and a DRNN neural network model;

[0026] The error and rate of change between the parameter control value and the parameter detection value are used as inputs to the fuzzy RBF neural network model-PID controller. The error and rate of change between the output of the fuzzy RBF neural network model-PID controller and the output of the NARX neural network model are used as inputs to the NARX neural network model-PID controller. The output of the NARX neural network model-PID controller and the parameter detection value are used as corresponding inputs to the NARX neural network model. The error and rate of change between the output of the NARX neural network model-PID controller and the output of the DRNN neural network model are used as inputs to the DRNN neural network model-PID controller. The output of the DRNN neural network model-PID controller and the parameter detection value are used as corresponding inputs to the DRNN neural network model. The output of the DRNN neural network model-PID controller is used as the output of the sprinkler irrigation controller to control the sprinkler irrigation device.

[0027] Furthermore, multiple moisture sensors are evenly distributed in each spray irrigation area.

[0028] The sprinkler irrigation control device of the present invention includes a memory and a processor. The memory stores a computer program, and when the computer program is executed by the processor, it implements the sprinkler irrigation control method of the claims.

[0029] The sprinkler irrigation control IoT system of the present invention includes a moisture sensing and control terminal 1-n, a sprinkler irrigation transmission terminal, a sprinkler irrigation cloud server, and a sprinkler irrigation application terminal. The sprinkler irrigation application terminal includes a sprinkler irrigation monitoring terminal and a sprinkler irrigation interactive APP. Each moisture sensing and control terminal interacts wirelessly with the sprinkler irrigation cloud server through the sprinkler irrigation transmission terminal. The sprinkler irrigation monitoring terminal and the sprinkler irrigation interactive APP are used to facilitate remote monitoring of sprinkler irrigation by users and to achieve precise adjustment of the sprinkler irrigation device.

[0030] Furthermore, the moisture sensing and control terminal includes a temperature sensor, a humidity sensor, a rainfall sensor, a wind speed sensor, a wind direction sensor, a carbon dioxide sensor, a moisture sensor, a light intensity sensor, an STM32 microprocessor, a camera, a LoRa communication module, and a sprinkler irrigation device. Each sensor is connected to the STM32 microprocessor through a corresponding signal conditioning circuit; the camera, LoRa communication module, and sprinkler irrigation device are each connected to the STM32 microprocessor; the moisture sensing and control terminal interacts with the sprinkler irrigation transmission terminal through a self-organizing communication network.

[0031] Furthermore, the sprinkler irrigation transmission end includes an STM32 microprocessor, as well as a LoRa communication module and a wireless communication module connected to the STM32 microprocessor respectively; the sprinkler irrigation transmission end constructs a self-organizing communication network with the moisture sensing and control end through the LoRa communication module; the wireless communication module realizes bidirectional data interaction between the sprinkler irrigation transmission end and the sprinkler irrigation cloud server.

[0032] Furthermore, the sprinkler irrigation monitoring terminal uses an industrial control computer. The sprinkler irrigation monitoring terminal and the sprinkler irrigation interactive APP exchange information with the sprinkler irrigation cloud server through a 5G network. The functions of the sprinkler irrigation monitoring terminal include sprinkler irrigation communication parameter setting, sprinkler irrigation data analysis, sprinkler irrigation data management, and sprinkler irrigation control model. The functions of the sprinkler irrigation interactive APP include sprinkler irrigation communication parameter setting, sprinkler irrigation data analysis, sprinkler irrigation data management, sprinkler irrigation alarm, and sprinkler irrigation process parameter acquisition and moisture monitoring.

[0033] Beneficial effects: Compared with the prior art, the present invention has the following significant advantages:

[0034] (1) The fuzzy RBF neural network model combines fuzzy theory with RBF neural network. After fuzzifying the input sprinkler irrigation parameter signal, the input sprinkler irrigation parameter signal is no longer just a specific number. It effectively preserves the sprinkler irrigation information. By adding a sprinkler irrigation parameter fuzzification layer in front of the input layer of the RBF neural network, that is, fuzzifying the input sprinkler irrigation signal and using it as the input sprinkler irrigation signal of the RBF neural network model, it can effectively overcome the problems that the RBF neural network model cannot process and describe fuzzy information, as well as the problems that fuzzy reasoning requires a lot of manual intervention and the low accuracy of sprinkler irrigation parameters.

[0035] (2) The fuzzy RBF neural network model-PID controller utilizes the characteristics of heuristic search of sprinkler irrigation parameters by fuzzy system and the fast inference speed of RBF neural network model to achieve high-precision tracking of water content of sprinkler irrigation object. By comparing and analyzing the control methods of fuzzy RBF neural network model-PID controller, fuzzy neural network model-PID controller and RBF neural network model-PID controller and tracking sprinkler irrigation water error, it is found that the fuzzy RBF neural network model-PID controller has a shorter response time, smaller steady-state error and more effective sprinkler irrigation control performance.

[0036] (3) When using a traditional PID controller, the traditional PID controller relies too heavily on the given proportional, integral, and derivative coefficients. In environments with changing loads, the traditional PID controller cannot adjust its parameters in a timely manner, thus affecting the control effect. Fuzzy control has the characteristic of strong reasoning ability. The combination of a fuzzy RBF neural network model and a PID controller can realize the self-tuning of PID parameters. However, the sprinkler irrigation control rules of the fuzzy neural network model rely too heavily on the subjective experience of experts. Although intuitive, they also have a certain degree of uncertainty, making it impossible to reasonably adjust the control rules during the control process and significantly reduce the steady-state error of the system. The fuzzy RBF neural network model-PID controller utilizes the characteristics of strong self-learning and adaptive capabilities of the RBF neural network model, which makes up for the lack of learning ability of fuzzy control for sprinkler irrigation parameters. Combining the two can realize the flexible online self-tuning function of PID controller parameters.

[0037] (4) The NARX neural network model is a global feedback dynamic neural network with external input sprinkler irrigation parameters, which is often used to describe nonlinear discrete systems. The NARX neural network model establishes a single system model with multiple inputs and multiple outputs, taking into account the direct coupling relationship between the input and output variables of the sprinkler irrigation system. It establishes the model and analyzes related problems from the perspective of the overall measured and controlled sprinkler irrigation system. The NARX neural network model method fully considers the influence of historical data of the input sprinkler irrigation information and output information on the current state of the water measurement and control system. This makes the model of the identified sprinkler irrigation measurement and control system more consistent with the actual process.

[0038] (5) The moisture detection module combines the dynamic network performance of the DRNN neural network model with feedback, the bidirectional prediction of water performance in the sprinkler irrigation process of the BiGRU neural network model, and the fuzzy inference and global feedback features of the fuzzy RBF neural network model-NARX neural network model, which improves the speed, accuracy, reliability and robustness of moisture parameter detection.

[0039] (6) The sprinkler irrigation controller includes a DRNN neural network model-PID controller, a NARX neural network model-PID controller, a fuzzy RBF neural network model-PID controller, a fuzzy RBF neural network model-NARX neural network model, and a fuzzy RBF neural network model-DRNN neural network model; through three-level intelligent PID control and two-level intelligent prediction of the controlled sprinkler irrigation parameters, and each level of control and prediction is nested on the basis of the previous level, the speed, accuracy, robustness and reliability of the water in the controlled sprinkler irrigation process are improved.

[0040] (7) The sprinkler irrigation control model takes into account the influence between adjacent sprinkler irrigation areas, which further improves the accuracy of water regulation. Attached Figure Description

[0041] Figure 1 This is a schematic diagram of the structure of the sprinkler irrigation control model in the embodiments of this application;

[0042] Figure 2 This is a schematic diagram of the moisture detection module in an embodiment of this application;

[0043] Figure 3 This is a schematic diagram of the structure of the sprinkler irrigation controller in the embodiments of this application;

[0044] Figure 4 This is a schematic diagram of the structure of the sprinkler irrigation control Internet of Things system in the embodiments of this application;

[0045] Figure 5 This is a schematic diagram of the structure of the moisture sensing and control terminal in the embodiments of this application;

[0046] Figure 6 This is a schematic diagram of the structure of the sprinkler irrigation transmission end in an embodiment of this application;

[0047] Figure 7 This is a schematic diagram of the sprinkler irrigation monitoring terminal software in the embodiments of this application. Detailed Implementation

[0048] The invention will now be further described with reference to the accompanying drawings.

[0049] This application provides a method for controlling sprinkler irrigation, wherein the sprinkler irrigation area includes adjacent sprinkler irrigation area 1, sprinkler irrigation area 2 and sprinkler irrigation area 3.

[0050] The spray irrigation control method includes:

[0051] The outputs of multiple moisture sensors in sprinkler irrigation area 1, sprinkler irrigation area 2, and sprinkler irrigation area 3 are acquired.

[0052] Constructing a sprinkler irrigation control model:

[0053] like Figure 1 The sprinkler irrigation control model includes NARX neural network models 1-3, DRNN neural network models 1-3, sprinkler irrigation controllers 1-3, moisture detection modules 1-3, fuzzy RBF neural network model-DRNN neural network model 1-2, and fuzzy RBF neural network model-NARX neural network model. Moisture detection module 1 is used to obtain the moisture detection value of sprinkler irrigation area 1 based on the outputs of multiple moisture sensors in sprinkler irrigation area 1, and the moisture detection value of sprinkler irrigation area 1 is used as the input of sprinkler irrigation controller 1. Moisture detection module 2 is used to obtain the moisture detection value of sprinkler irrigation area 2 based on the outputs of multiple moisture sensors in sprinkler irrigation area 2, and the moisture detection value of sprinkler irrigation area 2 is used as the input of sprinkler irrigation controller 2. Moisture detection module 3 is used to obtain the moisture detection value of sprinkler irrigation area 3 based on the outputs of multiple moisture sensors in sprinkler irrigation area 3, and the moisture detection value of sprinkler irrigation area 3 is used as the input of sprinkler irrigation controller 3.

[0054] The target moisture value of the sprinkler irrigation area 1 is used as the input of the NARX neural network model 1 and the moisture control value of the sprinkler irrigation controller 1, respectively. The sum of the output of the sprinkler irrigation controller 1 and the output of the NARX neural network model 1 is used as the corresponding input of the DRNN neural network model 1 and the corresponding input of the fuzzy RBF neural network model-DRNN neural network model 1, respectively. The output of the DRNN neural network model 1 is used as the input of the sprinkler irrigation device 1. The sprinkler irrigation device 1 is used to adjust the moisture of the sprinkler irrigation area 1.

[0055] The target moisture value of the sprinkler irrigation area 2 is used as the input of NARX neural network model 2 and the corresponding input of fuzzy RBF neural network model-DRNN neural network model 1, respectively. The output of fuzzy RBF neural network model-DRNN neural network model 1 is used as the moisture control value of sprinkler irrigation controller 2. The sum of the output of sprinkler irrigation controller 2 and the output of NARX neural network model 2 is used as the corresponding input of DRNN neural network model 2 and fuzzy RBF neural network model-DRNN neural network model 2, respectively. The output of DRNN neural network model 2 is used as the input of sprinkler irrigation device 2. Sprinkler irrigation device 2 is used to adjust the moisture of sprinkler irrigation area 2.

[0056] The target moisture value of the sprinkler irrigation area 3 is used as the input of NARX neural network model 3 and the corresponding input of fuzzy RBF neural network model-DRNN neural network model 2, respectively. The output of fuzzy RBF neural network model-DRNN neural network model 2 is used as the moisture control value of sprinkler irrigation controller 3. The sum of the output of sprinkler irrigation controller 3 and the output of NARX neural network model 3 is used as the corresponding input of DRNN neural network model 3. The output of DRNN neural network model 3 is used as the input of sprinkler irrigation device 3. Sprinkler irrigation device 3 is used to adjust the moisture of sprinkler irrigation area 3.

[0057] The outputs of DRNN neural network model 1, DRNN neural network model 2, and DRNN neural network model 3 are used as the corresponding inputs of the fuzzy RBF neural network model-NARX neural network model, respectively. The outputs of moisture detection module 1, moisture detection module 2, and moisture detection module 3 are used as the corresponding inputs of the fuzzy RBF neural network model-NARX neural network model, respectively. The moisture disturbance amount of the sprinkler irrigation area 1 output by the fuzzy RBF neural network model-NARX neural network model is used as the corresponding input of DRNN neural network model 1, the moisture disturbance amount of the sprinkler irrigation area 2 output is used as the corresponding input of DRNN neural network model 2, and the moisture disturbance amount of the sprinkler irrigation area 3 output is used as the corresponding input of DRNN neural network model 3.

[0058] Multiple moisture sensors are evenly distributed in the sprinkler irrigation area 1, multiple moisture sensors are evenly distributed in the sprinkler irrigation area 2, and multiple moisture sensors are evenly distributed in the sprinkler irrigation area 3.

[0059] like Figure 2 The moisture detection module includes DRNN neural network models 1-3, BiGRU neural network models 1-3, and fuzzy RBF neural network model-NARX neural network model;

[0060] The outputs of multiple moisture sensors serve as the corresponding inputs to DRNN neural network model 1, DRNN neural network model 2, and DRNN neural network model 3, respectively. The output of DRNN neural network model 1 serves as the corresponding input to BiGRU neural network model 1 and fuzzy RBF neural network model-NARX neural network model, respectively. The output of DRNN neural network model 2 serves as the corresponding input to BiGRU neural network model 2 and fuzzy RBF neural network model-NARX neural network model, respectively. The outputs of BiGRU neural network model 1, BiGRU neural network model 2, and BiGRU neural network model 3 serve as the corresponding input to fuzzy RBF neural network model-NARX neural network model, respectively. The output of fuzzy RBF neural network model-NARX neural network model is used as the moisture detection value of the sprinkler irrigation area.

[0061] like Figure 3 The sprinkler irrigation controller includes a fuzzy RBF neural network model-PID controller, a NARX neural network model-PID controller, a DRNN neural network model-PID controller, a NARX neural network model, and a DRNN neural network model;

[0062] The parameter control value and parameter detection value are used as inputs to the sprinkler irrigation controller. The error and error rate of change between the parameter control value and the parameter detection value are used as inputs to the fuzzy RBF neural network model-PID controller. The error and error rate of change between the output of the fuzzy RBF neural network model-PID controller and the output of the NARX neural network model are used as inputs to the NARX neural network model-PID controller. The output of the NARX neural network model-PID controller and the parameter detection value are used as corresponding inputs to the NARX neural network model. The error and error rate of change between the output of the NARX neural network model-PID controller and the output of the DRNN neural network model are used as inputs to the DRNN neural network model-PID controller. The output of the DRNN neural network model-PID controller and the parameter detection value are used as corresponding inputs to the DRNN neural network model. The output of the DRNN neural network model-PID controller is used as the output of the sprinkler irrigation controller to control the sprinkler irrigation device.

[0063] I. Moisture Detection Module Design

[0064] 1) DRNN Neural Network Model

[0065] The DRNN neural network model is a locally recurrent neural network simplified from a fully recurrent neural network. It possesses a dynamic feedback network, capable of reflecting the dynamic characteristics of sprinkler irrigation systems. Furthermore, it trains the network based on the structure of a BP neural network, thus enabling the nonlinear, time-varying prediction system of sprinkler irrigation to adapt to time-varying characteristics. The hidden layer units of the DRNN neural network model are self-recursive, but the connection weights between hidden layer nodes are reduced, significantly decreasing the computational load. It also retains the dynamic learning characteristics of general recurrent neural networks. Considering the characteristics of sprinkler irrigation control systems, such as multivariable and time-varying characteristics, harsh working conditions, and numerous random disturbances, as well as the control characteristics of diagonal recurrent neural networks, a 3-layer feedforward network architecture of the DRNN neural network model is chosen. In this model, any neuron in the hidden layer (i.e., the regression layer) only receives feedback from its own output. In the DRNN neural network model, I... i (k) is the i-th input of the first layer, i.e., the input layer. It is the weight vector from the input layer to the hidden layer. It is the weight vector of the hidden layer, i.e., the regression layer, of the network. X is the weight vector from the hidden layer to the output layer of the network. j (k) is the output of the j-th neuron in the regression layer of the network, u(k) is the output of the 3rd layer (output layer) in the DRNN neural network model, and S j (k) is the sum of activations of the j-th regression neuron in the second hidden layer of the DRNN neural network model. The specific calculation formula is as follows:

[0066]

[0067] 2) BiGRU Neural Network Model

[0068] The BiGRU neural network model consists of a forward GRU neural network model and a backward GRU neural network model. Each GRU neural network unit contains two gate structures: a reset gate and an update gate. The reset gate's function is to forget the hidden layer unit h from the previous time step. t-1 The information is obtained from the previous hidden state, while the update gate controls the balance between the previous hidden state and the current input information. The specific derivation formula is as follows:

[0069] r t =σ(W r ·[h t-1 ,x t ]+b r (2)

[0070] z t =σ(W z ·[h t-1 ,x t ]+b z(3)

[0071]

[0072]

[0073] In the formula: σ is the sigmoid activation function; W and b are the parameter matrices of the control gate; x t The input data at time t; h t The hidden state to be passed to the next node; For candidate hidden states; r t To reset the door, discard irrelevant information by controlling the input from the previous hidden state; t The update gate, used for forgetting and selective memory, updates information to a new state h. t or candidate state The BiGRU neural network model consists of two GRU neural network model units operating in opposite directions. These networks process the input sprinkler irrigation parameter sequence sequentially in the forward and backward directions of time, respectively. The outputs of the GRU neural network model at each time step are concatenated to form the final output layer of the sprinkler irrigation parameters. One network propagates the sprinkler irrigation parameter information forward, while the other propagates it in the opposite direction. When past sprinkler irrigation parameter information influences future data, the future information simultaneously maintains a correlation with past information, exhibiting good predictive and recognition capabilities. It can also capture the time-scale dependencies of various sprinkler irrigation parameters and provide additional contextual information. This allows the BiGRU neural network model to effectively learn the forward and backward time-series information in the input sprinkler irrigation parameter sequence, improving the accuracy of predicting the input sprinkler irrigation information.

[0074] 3) Fuzzy RBF Neural Network Model - NARX Neural Network Model

[0075] The fuzzy RBF neural network model-NARX neural network model is a concatenation of the fuzzy RBF neural network model and the NARX neural network model. The fuzzy RBF neural network model consists of an input layer, a fuzzification layer, a fuzzy inference layer, and an output layer. The signal transmission and structural functions of the fuzzy RBF neural network model are as follows:

[0076] Layer 1: Input Layer. This layer does not process the input sprinkler irrigation data; instead, it directly passes the input sprinkler irrigation data to the next layer. Its nodes are connected to all nodes in the next layer. The input and output of each node i in the input layer are the input sprinkler irrigation parameter vectors directly connected to different nodes in the input layer. The connection function of the i-th node in the input layer is expressed as:

[0077] f1(i)=x i,i=1,2,…n(6)

[0078] Layer 2: Fuzzification Layer, also known as the membership function layer. The output of each node in the membership function layer can be considered a membership function. Since this is an RBF neural network model, the commonly used Gaussian function is used as the membership function. The connection function between the i-th node of the input layer and the j-th node of the fuzzification layer is expressed as:

[0079]

[0080] In the formula: c ij Let b be the mean of the Gaussian function of the j-th fuzzy set of the i-th input signal. j Let be the mean of the Gaussian function of the j-th fuzzy set for each input signal, where j = 1, 2, ..., m.

[0081] Layer 3: Fuzzy Inference Layer, also known as the Rule Layer. The Rule Layer connects with the Membership Function Layer to complete fuzzy rule matching. Fuzzy calculations are performed between the sprinkler irrigation parameter nodes in the Membership Function Layer to obtain the corresponding ignition intensity. The number of nodes in the Rule Layer is equal to the number of nodes in the Membership Function Layer; therefore, the output of the j-th node in the Rule Layer is the product of all input sprinkler irrigation signals for that node, i.e.:

[0082]

[0083] In the formula: N i Input the number of membership functions of the sprinkler irrigation parameters for the i-th node in the input layer.

[0084] Layer 4: Output Layer. The output of each node in the output layer is the weighted sum of all input sprinkler irrigation parameter signals for that node, i.e.:

[0085]

[0086] In the formula: l is the number of output layer nodes, and W is the connection weight matrix between the output layer nodes and the rule layer nodes.

[0087] The NARX neural network model consists of an input layer, hidden layers, and an output layer. The input sprinkler irrigation parameters of the NARX neural network model simultaneously include the current input sprinkler irrigation parameter values ​​and the output values ​​from historical times, making it a neural network with memory capabilities. Therefore, using the NARX neural network model for model identification can effectively utilize the historical input and output data of the input sprinkler irrigation parameters and can also take into account the coupling between the input and output variables. Its mathematical expression is:

[0088] y(t)=f[x(t),…,x(t-d1),y(t-1),…,y(t-d2)](10)

[0089] Where x(t) and y(t) are the input and output parameters of the NARX neural network model at time t, respectively; d1 and d2 are the input and output delay orders of the NARX neural network model, respectively; and f[] is the nonlinear function fitted by the NARX neural network model. Before applying the NARX neural network model, it is necessary to determine the network parameters such as the number of input and output nodes, the input and output delay orders, and the number of hidden layer neurons. The number of nodes in the input and output layers should be determined according to the actual research object, and the number of hidden layer neurons depends on the following empirical formula:

[0090]

[0091] Where p is the number of hidden layer neurons, m is the number of input layer nodes, and n is the number of output layer nodes. This determines the range of the number of hidden layer neurons, and then the best-performing value within that range can be selected.

[0092] II. Design of Sprinkler Irrigation Controller

[0093] The sprinkler irrigation controller includes a DRNN neural network model-PID controller, a NARX neural network model-PID controller, a fuzzy RBF neural network model-PID controller, a NARX neural network model, and a DRNN neural network model. The design methods of the DRNN neural network model, NARX neural network model, fuzzy RBF neural network model, and fuzzy RBF neural network model-NARX neural network model refer to the design methods of the moisture detection module.

[0094] 4) DRNN Neural Network Model - PID Controller

[0095] DRNN Neural Network Model - Output K of the DRNN Neural Network Model for PID Controller P K I and K D The proportional, integral, and derivative coefficients of a PID controller are used. PID closed-loop control mainly consists of proportional (P), integral (I), and derivative (D). Effective water control for sprinkler irrigation is achieved by calculating the control quantity through proportional, integral, and derivative coefficients. The fundamental principle of a PID controller is proportional control. While integral control can effectively reduce steady-state water error, it is highly likely to increase overshoot. Derivative control can accelerate the response speed of a large-inertia water system and effectively reduce overshoot. The relationship between the input and output u(t) of a PID controller is as follows:

[0096]

[0097] Where u(t) represents the output; e(t) represents the input; KP Represents the proportionality coefficient; K I K represents the integral coefficient; D Represents the differential coefficient.

[0098] 5) NARX Neural Network Model - PID Controller

[0099] NARX Neural Network Model - Output K of the NARX Neural Network Model for PID Controller P K I and K D The proportional, integral, and derivative coefficients of a PID controller are used. The design process of a PID controller follows the design method of a DRNN neural network model for PID controllers.

[0100] 6) Fuzzy RBF Neural Network Model - PID Controller

[0101] Fuzzy RBF Neural Network Model - Output K of the Fuzzy RBF Neural Network Model for PID Controller P K I and K D The proportional, integral, and derivative coefficients of a PID controller are used. The design process of a PID controller follows the design method of a DRNN neural network model for PID controllers.

[0102] III. Design of Sprinkler Irrigation Control Model

[0103] The sprinkler irrigation control model includes a NARX neural network model, a DRNN neural network model, a sprinkler irrigation controller, a moisture detection module, a fuzzy RBF neural network model-DRNN neural network model, and a fuzzy RBF neural network model-NARX neural network model; the fuzzy RBF neural network model-DRNN neural network model is a cascaded fuzzy RBF neural network model and a DRNN neural network model; the design method of each component model in the sprinkler irrigation control model is the same as that of the moisture detection module.

[0104] This application also provides a sprinkler irrigation control device, including a memory and a processor. The memory stores a computer program, which, when executed by the processor, implements the sprinkler irrigation control method. This sprinkler irrigation control device has the advantages of the sprinkler irrigation control method.

[0105] This application also provides an IoT system for controlling sprinkler irrigation, including a moisture sensing and control terminal 1-n, a sprinkler irrigation transmission terminal, a sprinkler irrigation cloud server, and a sprinkler irrigation application terminal. The sprinkler irrigation application terminal includes a sprinkler irrigation monitoring terminal and a sprinkler irrigation interactive APP, which exchange information with the sprinkler irrigation cloud server through a 5G network. Each moisture sensing and control terminal wirelessly exchanges data with the sprinkler irrigation cloud server through the sprinkler irrigation transmission terminal. The sprinkler irrigation monitoring terminal and the sprinkler irrigation interactive APP are used to facilitate remote monitoring of sprinkler irrigation by users and to achieve precise adjustment of the sprinkler irrigation device.

[0106] The number of water sensing and control terminals set up in the sprinkler irrigation area should be reasonably determined and arranged according to the area area; no specific limit is set in this case.

[0107] like Figure 5 The moisture sensing and control unit includes a temperature sensor, humidity sensor, rainfall sensor, wind speed sensor, wind direction sensor, carbon dioxide sensor, moisture sensor, illuminance sensor, STM32 microprocessor, camera, LoRa communication module, and sprinkler irrigation device. Each sensor is connected to the STM32 microprocessor through a corresponding signal conditioning circuit. The camera, LoRa communication module, and sprinkler irrigation device are all connected to the STM32 microprocessor. The camera acquires video information of the controlled object (i.e., the sprinkler irrigation device), and the sprinkler irrigation device adjusts the moisture content of the corresponding sprinkler irrigation area. A self-organizing communication network based on numerous LoRa communication modules is used to collect sprinkler irrigation process parameters and perform moisture control. The moisture sensing and control unit interacts with the sprinkler irrigation transmission unit through the self-organizing communication network.

[0108] like Figure 6 The sprinkler irrigation transmission end includes an STM32 microprocessor, as well as a LoRa communication module and a wireless communication module connected to the STM32 microprocessor respectively; the sprinkler irrigation transmission end constructs a self-organizing communication network with the water sensing and control end through the LoRa communication module; the wireless communication module realizes bidirectional data interaction between the sprinkler irrigation transmission end and the sprinkler irrigation cloud server.

[0109] The sprinkler irrigation monitoring terminal uses an industrial control computer. Its functions include setting sprinkler irrigation communication parameters, analyzing sprinkler irrigation data, managing sprinkler irrigation data, and developing a sprinkler irrigation control model. Figure 7 As shown, the functions of the sprinkler irrigation interactive APP include sprinkler irrigation communication parameter settings, sprinkler irrigation data analysis, sprinkler irrigation data management, sprinkler irrigation alarms, and sprinkler irrigation process parameter acquisition and moisture monitoring.

[0110] Multiple moisture sensing and control terminals integrate data from temperature, humidity, carbon dioxide concentration, illuminance, wind speed, wind direction, rainfall, and moisture sensors, as well as data collected by cameras. This data is then uploaded in real time to the sprinkler irrigation transmission terminal via a LoRa communication module. The sprinkler irrigation transmission terminal transmits this data to the sprinkler irrigation cloud server for storage, updating, and management via a wireless communication module. The sprinkler irrigation application terminal sends data acquisition and moisture control commands to the sprinkler irrigation transmission terminal through the sprinkler irrigation cloud server. The sprinkler irrigation transmission terminal receives these commands and sends them back to the moisture sensing and control terminal, thus enabling remote monitoring of data acquisition and moisture levels.

[0111] This invention utilizes the Internet of Things and a constructed sprinkler irrigation control model to achieve digital control and networked management of sprinkler irrigation. It enables remote information collection and precise adjustment of the sprinkler irrigation device, while also facilitating online monitoring of the sprinkler irrigation system by users.

Claims

1. A method for controlling sprinkler irrigation, characterized in that, The sprinkler irrigation area includes adjacent sprinkler irrigation area 1 and sprinkler irrigation area 2, and the sprinkler irrigation control method includes: Acquire the outputs of multiple moisture sensors in sprinkler irrigation area 1 and multiple moisture sensors in sprinkler irrigation area 2; Constructing a sprinkler irrigation control model: The sprinkler irrigation control model includes NARX neural network models 1-2, DRNN neural network models 1-2, sprinkler irrigation controllers 1-2, moisture detection modules 1-2, fuzzy RBF neural network model-DRNN neural network model 1, and fuzzy RBF neural network model-NARX neural network model; wherein, moisture detection module 1 is used to obtain the moisture detection value of sprinkler irrigation area 1 based on the output of multiple moisture sensors in sprinkler irrigation area 1, and the moisture detection value of sprinkler irrigation area 1 is used as the input of sprinkler irrigation controller 1; moisture detection module 2 is used to obtain the moisture detection value of sprinkler irrigation area 2 based on the output of multiple moisture sensors in sprinkler irrigation area 2, and the moisture detection value of sprinkler irrigation area 2 is used as the input of sprinkler irrigation controller 2; The target moisture value of the sprinkler irrigation area 1 is used as the input of the NARX neural network model 1 and the moisture control value of the sprinkler irrigation controller 1, respectively. The sum of the output of the sprinkler irrigation controller 1 and the output of the NARX neural network model 1 is used as the corresponding input of the DRNN neural network model 1 and the corresponding input of the fuzzy RBF neural network model-DRNN neural network model 1, respectively. The output of the DRNN neural network model 1 is used as the input of the sprinkler irrigation device 1. The sprinkler irrigation device 1 is used to adjust the moisture of the sprinkler irrigation area 1. The target moisture value of the sprinkler irrigation area 2 is used as the input of NARX neural network model 2 and the corresponding input of fuzzy RBF neural network model-DRNN neural network model 1, respectively. The output of fuzzy RBF neural network model-DRNN neural network model 1 is used as the moisture control value of sprinkler irrigation controller 2. The sum of the output of sprinkler irrigation controller 2 and the output of NARX neural network model 2 is used as the corresponding input of DRNN neural network model 2. The output of DRNN neural network model 2 is used as the input of sprinkler irrigation device 2. Sprinkler irrigation device 2 is used to adjust the moisture of sprinkler irrigation area 2. The outputs of DRNN neural network model 1 and DRNN neural network model 2 are used as the corresponding inputs of the fuzzy RBF neural network model-NARX neural network model, respectively. The outputs of moisture detection module 1 and moisture detection module 2 are used as the corresponding inputs of the fuzzy RBF neural network model-NARX neural network model, respectively. The moisture disturbance amount of the sprinkler irrigation area 1 output by the fuzzy RBF neural network model-NARX neural network model is used as the corresponding input of DRNN neural network model 1, and the moisture disturbance amount of the sprinkler irrigation area 2 output is used as the corresponding input of DRNN neural network model 2. The sprinkler irrigation area also includes a sprinkler irrigation area 3 adjacent to sprinkler irrigation area 2, and the sprinkler irrigation control method further includes: Acquire the outputs of three moisture sensors in the sprinkler irrigation area; The sprinkler irrigation control model also includes NARX neural network model 3, DRNN neural network model 3, sprinkler irrigation controller 3, moisture detection module 3 and fuzzy RBF neural network model-DRNN neural network model 2; the moisture detection module 3 is used to obtain the moisture detection value of the sprinkler irrigation area 3 based on the output of multiple moisture sensors in the sprinkler irrigation area 3, and the moisture detection value of the sprinkler irrigation area 3 is used as the input of the sprinkler irrigation controller 3. The sum of the outputs of the sprinkler irrigation controller 2 and the NARX neural network model 2 is also used as the corresponding input of the fuzzy RBF neural network model - DRNN neural network model 2; The target moisture value of the sprinkler irrigation area 3 is used as the input of NARX neural network model 3 and the corresponding input of fuzzy RBF neural network model-DRNN neural network model 2, respectively. The output of fuzzy RBF neural network model-DRNN neural network model 2 is used as the moisture control value of sprinkler irrigation controller 3. The sum of the output of sprinkler irrigation controller 3 and the output of NARX neural network model 3 is used as the corresponding input of DRNN neural network model 3. The output of DRNN neural network model 3 is used as the input of sprinkler irrigation device 3. Sprinkler irrigation device 3 is used to adjust the moisture of sprinkler irrigation area 3. The output of DRNN neural network model 3 serves as the corresponding input of the fuzzy RBF neural network model-NARX neural network model, the output of moisture detection module 3 serves as the corresponding input of the fuzzy RBF neural network model-NARX neural network model, and the moisture disturbance amount of the sprinkler irrigation area 3 output by the fuzzy RBF neural network model-NARX neural network model serves as the corresponding input of DRNN neural network model 3. The sprinkler irrigation controller includes a fuzzy RBF neural network model-PID controller, a NARX neural network model-PID controller, a DRNN neural network model-PID controller, a NARX neural network model, and a DRNN neural network model; The error and rate of change between the parameter control value and the parameter detection value are used as inputs to the fuzzy RBF neural network model-PID controller. The error and rate of change between the output of the fuzzy RBF neural network model-PID controller and the output of the NARX neural network model are used as inputs to the NARX neural network model-PID controller. The output of the NARX neural network model-PID controller and the parameter detection value are used as corresponding inputs to the NARX neural network model. The error and rate of change between the output of the NARX neural network model-PID controller and the output of the DRNN neural network model are used as inputs to the DRNN neural network model-PID controller. The output of the DRNN neural network model-PID controller and the parameter detection value are used as corresponding inputs to the DRNN neural network model. The output of the DRNN neural network model-PID controller is used as the output of the sprinkler irrigation controller to control the sprinkler irrigation device.

2. The sprinkler irrigation control method according to claim 1, characterized in that, The moisture detection module includes DRNN neural network models 1-3, BiGRU neural network models 1-3, and fuzzy RBF neural network model-NARX neural network model; The outputs of multiple moisture sensors serve as the corresponding inputs to DRNN neural network model 1, DRNN neural network model 2, and DRNN neural network model 3, respectively. The output of DRNN neural network model 1 serves as the corresponding input to BiGRU neural network model 1 and fuzzy RBF neural network model-NARX neural network model, respectively. The output of DRNN neural network model 2 serves as the corresponding input to BiGRU neural network model 2 and fuzzy RBF neural network model-NARX neural network model, respectively. The outputs of BiGRU neural network model 1, BiGRU neural network model 2, and BiGRU neural network model 3 serve as the corresponding input to fuzzy RBF neural network model-NARX neural network model, respectively. The output of fuzzy RBF neural network model-NARX neural network model is used as the moisture detection value of the sprinkler irrigation area.

3. The sprinkler irrigation control method according to claim 1, characterized in that, Multiple moisture sensors are evenly distributed in each sprinkler irrigation area.

4. A sprinkler irrigation control device, comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the computer program is executed by a processor, it implements the sprinkler irrigation control method according to any one of claims 1 to 3.

Images