Intelligent moisture detection and control method, Internet of Things system and device thereof

Abstract

The invention discloses an intelligent moisture detection and control method, system and device, a plurality of moisture sensors, temperature sensors and humidity sensors are arranged, and a moisture prediction value is output after the moisture is processed by a moisture detection module; a water pump driving device, an electric valve and a flow sensor are arranged, control instructions of the water pump driving device and the electric valve are output through a moisture control module according to a moisture predicted value and a moisture target value, and the moisture value of a detected object is adjusted in real time to meet the moisture target value. Collected data is transmitted to a moisture cloud platform server, stored, processed and then sent to a moisture monitoring platform, and the moisture monitoring platform dynamically displays, inquires, monitors and issues a moisture adjusting instruction. Compared with the prior art, the method has the advantages that the accurate detection result can be quickly and effectively obtained, the moisture of the substance is accurately controlled, and the method is of great significance in improving the product quality.

Description

Technical Field

[0001] The present invention relates to the field of intelligent moisture detection and control automation technology, and in particular to an intelligent moisture detection and control method, an Internet of Things system and a device thereof. Background Art

[0002] The moisture content of a substance is one of the important detection and control parameters in the industrial and agricultural production process in many fields such as papermaking, tobacco, food, oil, and agricultural products. The moisture content is detected by a moisture sensor, but the sensor is easily disturbed by environmental parameters during the detection process, which greatly affects its measurement accuracy, resulting in the inability to accurately detect the moisture content of the substance. At the same time, in the process of regulating the moisture content of the substance, the moisture control accuracy is low due to the simple control algorithm, and the moisture content of the substance cannot be accurately controlled. This has a great impact on the later production in the fields of papermaking, tobacco, food, oil, and agricultural products.

[0003] The present invention solves the problem of the nonlinearity of environmental sensors and the coupling influence of environmental parameters on the moisture content of materials by designing a moisture detection module, thereby achieving rapid and effective acquisition of accurate detection results. The moisture content of materials is precisely controlled by designing a moisture control module, which is of great significance to improving product quality. Summary of the invention

[0004] Purpose of the invention: In view of the problems in the background technology, the present invention discloses an intelligent moisture detection and control method, an Internet of Things system and a device thereof, which can achieve fast and effective acquisition of accurate detection results and precise control of the moisture content of materials, which is of great significance to improving product quality.

[0005] Technical solution: The present invention discloses an intelligent moisture detection and control method, comprising the following steps:

[0006] Step 1: Set up multiple moisture sensors, multiple temperature sensors and multiple humidity sensors in the monitoring environment;

[0007] Step 2: Construct a moisture detection module, process the collected values ​​of multiple moisture sensors, multiple temperature sensors and multiple humidity sensors through the moisture detection module and output the moisture prediction value;

[0008] The moisture detection module includes GWO's NARX neural network model-GWO's BiLSTM neural network model 1-3, GWO's fuzzy recursive neural network model-GWO's NARX neural network model 1 and GWO's fuzzy recursive neural network model-GWO's BiLSTM neural network model 1;

[0009] The outputs of multiple moisture sensors, multiple temperature sensors and multiple humidity sensors are respectively used as inputs of GWO's NARX neural network model-GWO's BiLSTM neural network model 1-3, the outputs of multiple humidity sensors and the outputs of GWO's NARX neural network model-GWO's BiLSTM neural network model 1-3 are used as corresponding inputs of GWO's fuzzy recursive neural network model-GWO's NARX neural network model 1, the outputs of multiple moisture sensors, the outputs of GWO's NARX neural network model-GWO's BiLSTM neural network model 1 and the outputs of GWO's fuzzy recursive neural network model-GWO's NARX neural network model 1 are used as corresponding inputs of GWO's fuzzy recursive neural network model-GWO's BiLSTM neural network model 1, and the outputs of GWO's fuzzy recursive neural network model-GWO's BiLSTM neural network model 1 are used as the moisture prediction value of the detected object;

[0010] Step 3: Set up a water pump driving device, an electric valve and a flow sensor 1-2, wherein the flow sensor 1-2 measures the flow rate of water driven and output by the water pump driving device and the flow rate of water controlled and output by the electric valve respectively;

[0011] Step 4: According to the moisture prediction value and the moisture target value, the moisture control module controls the control instructions of the water pump drive device and the electric valve, and adjusts the moisture value of the detected object in real time to make it meet the moisture target value.

[0012] Further, the moisture control module includes a NARX neural network model of GWO-a BiLSTM neural network model of GWO 4-5, a fuzzy recursive neural network model of GWO-a NARX neural network model of GWO 2, a fuzzy recursive neural network model of GWO-a BiLSTM neural network model of GWO 2-4, a moisture detection module and a PID controller 1-3;

[0013] The moisture target value is used as the input of GWO's NARX neural network model - GWO's BiLSTM neural network model 4, the error and error change rate between the moisture target value and the output of GWO's NARX neural network model - GWO's BiLSTM neural network model 5 are used as the input of GWO's fuzzy recurrent neural network model - GWO's NARX neural network model 2 and the input of PID controller 1, and the output of PID controller 1 and the output of GWO's NARX neural network model - GWO's BiLSTM neural network model 4 are used as GWO's fuzzy recurrent neural network model - GWO's BiLSTM neural network model 2 The output of GWO's fuzzy recursive neural network model - GWO's BiLSTM neural network model 2 is used as the input of the water pump drive device and the input of GWO's fuzzy recursive neural network model - GWO's BiLSTM neural network model 3. The error and error change rate of the water flow output by the water pump drive device and the moisture prediction value output by the moisture detection module are used as the input of PID controller 2 and GWO's fuzzy recursive neural network model - GWO's NARX neural network model 2. The output of PID controller 2 is used as the input of GWO's fuzzy recursive neural network model - GWO's BiLSTM neural network model 3. The output of BiLSTM neural network model 3 of WO-GWO is used as the input of BiLSTM neural network model 4 of GWO. The error and error change rate between the output of BiLSTM neural network model 3 of GWO-GWO and the output of flow sensor 2 are used as the input of NARX neural network model 2 of WO-GWO and the input of PID controller 3. The output of PID controller 3 is used as the input of BiLSTM neural network model 4 of GWO-GWO. The output of WO's fuzzy recursive neural network model - GWO's BiLSTM neural network model 4 is used as the input of GWO's fuzzy recursive neural network model - GWO's BiLSTM neural network model 2-3 and the input of the electric valve. The electric valve outputs the water flow to intelligently adjust the moisture of the moisture regulation object. The output of the moisture sensor group is used as the input of the moisture detection module. The output of the flow sensor 1-2 and the output of the moisture detection module are used as the input of GWO's NARX neural network model - GWO's BiLSTM neural network model 5. GWO's fuzzy recursive neural network model - GWO's NARX neural network model 2 outputs 3 groups of k p , k i and k d As k of PID controller 1-3 respectively p , k i and k d ;

[0014] The moisture sensor group includes a plurality of moisture sensors, a plurality of temperature sensors and a plurality of humidity sensors arranged in the monitoring environment.

[0015] Furthermore, the outputs of the PID controllers 1-3 are:

[0016]

[0017] Where u(t) is the control signal output by the PID controller at time t, k p is the proportionality coefficient, k i is the integration coefficient, k d is the differential coefficient, e(t) is the error between the controller input signal and the expected value, is the derivative of the error, that is, the rate of change of the error; k p The larger the ratio, the more obvious the link, the faster the water regulation response, and the more stable the water regulation. p The increase of moisture parameter error increases; k i The smaller the k is, the more obvious the effect on the integral link is, which makes the steady-state error of moisture parameter adjustment continuously reduce, affects the response speed of moisture parameter adjustment to a certain extent, and prolongs the moisture parameter adjustment time; d The larger the moisture parameter overshoot is, the more obvious the suppression effect is, making the attenuation effect of moisture parameter regulation oscillation smaller. d Too large a value will slow down the response speed of moisture parameter adjustment.

[0018] Furthermore, the GWO's NARX neural network model - GWO's BiLSTM neural network model 1-5 are all GWO's NARX neural network model and GWO's BiLSTM neural network model connected in series, GWO's NARX neural network model optimizes GWO's NARX neural network model parameters, and GWO's BiLSTM neural network model optimizes GWO's BiLSTM neural network model parameters.

[0019] Further, the fuzzy recursive neural network model of GWO-NARX neural network model of GWO 1-2 are both fuzzy recursive neural network models of GWO and NARX neural network models of GWO connected in series, the fuzzy recursive neural network model of GWO is the GWO optimized fuzzy recursive neural network model parameters, and the NARX neural network model of GWO is the GWO optimized NARX neural network model parameters;

[0020] The fuzzy recursive neural network model of GWO - the BiLSTM neural network model of GWO 1-4 are all the fuzzy recursive neural network model of GWO and the BiLSTM neural network model of GWO connected in series. The fuzzy recursive neural network model of GWO is the GWO optimized fuzzy recursive neural network model parameters, and the BiLSTM neural network model of GWO is the GWO optimized BiLSTM neural network model parameters.

[0021] The present invention also discloses an Internet of Things system for moisture detection and control, comprising a moisture measuring and controlling device, a moisture cloud platform server and a moisture monitoring platform; the moisture measuring and controlling device collects moisture, temperature, humidity, wind speed and illumination, and transmits the collected data to the moisture cloud platform server through a 4G communication module, the moisture cloud platform server processes and stores the received data, and sends the data to the moisture monitoring platform at the same time, the moisture monitoring platform dynamically displays, queries, monitors and issues moisture adjustment instructions to the moisture cloud platform server for the data of the moisture measuring and controlling device, and the moisture cloud platform server transmits the instructions to the moisture measuring and controlling device; the moisture cloud platform server executes the process of the above-mentioned intelligent moisture detection and control method when processing and storing the received data.

[0022] Preferably, the moisture monitoring platform includes a moisture monitoring terminal and a moisture monitoring APP terminal; the moisture monitoring terminal and the moisture monitoring APP terminal both dynamically display, query, monitor the data of the moisture measuring and controlling device and issue moisture adjustment instructions to the moisture cloud platform server.

[0023] The present invention also discloses an Internet of Things device for moisture detection and control, comprising a memory and a processor, wherein a computer program is stored in the memory, and when the computer program is executed by the processor, the process of the above-mentioned intelligent moisture detection and control method is implemented.

[0024] Compared with the prior art, the present invention has the following advantages:

[0025] 1. The NARX neural network model closed-loop time-delay network model designed by the present invention provides the moisture parameter prediction value as a new response input to the NARX neural network to dynamically predict the moisture parameter time series, thereby realizing the memory performance of moisture parameter prediction. The NARX neural network model is a dynamic prediction neural network, and its structure can effectively realize the discrete nonlinear autoregressive system prediction with external input, thereby improving the dynamics, robustness and accuracy of moisture parameter prediction.

[0026] 2. In the BiLSTM neural network model of the GWO of the present invention, the bidirectional propagation of moisture parameters can intercept the positive and negative moisture parameter characteristics due to the longer propagation path. When the moisture parameters are forwardly transmitted in the model, the model will record the forward transmitted moisture parameter characteristic information, and when the moisture parameters are reversely propagated, the model will record the reverse characteristic information of the moisture parameters.

[0027] 3. The fuzzy recursive neural network model designed by the present invention is a product of the combination of fuzzy theory and neural network. It has the fuzzy reasoning ability of the moisture parameter fuzzy system and the ability of the neural network to adapt to the changes in moisture parameters. The model can not only be automatically updated, but also continuously correct the membership function of the fuzzy subset. The recursive neural network increases the ability of the network itself to process the dynamic information of moisture parameters by adding some additional internal feedback channels to the fuzzy recursive neural network model. It can use the internal state of the network to feedback the changes in moisture parameters to describe the nonlinear dynamic behavior of the moisture parameter system, and can more directly reflect the dynamic characteristics of the moisture parameter system, that is, add self-feedback connections to the rule layer of the forward fuzzy network, and record the historical information of moisture parameters of the rule activation intensity. Tests show that the fuzzy recursive neural network model is applied to the dynamic modeling of moisture parameters, the test moisture parameter error is small, and the moisture parameter prediction accuracy is high.

[0028] 4. The Grey Wolf Optimization Algorithm (GWO) of the present invention is an intelligent algorithm for global optimization search, which has the advantages of fast convergence speed, few parameters, simple and easy implementation, and has adaptive adjustment factors and information feedback mechanism. GWO optimizes the parameters of the NARX neural network model, the BiLSTM neural network model of GWO and the fuzzy recursive neural network model, and improves the accuracy, robustness and rapidity of the predicted moisture parameters of the three models. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 It is a moisture detection module of the present invention;

[0030] Figure 2 The moisture control module of the present invention;

[0031] Figure 3 The Internet of Things system for moisture detection and control of the present invention;

[0032] Figure 4 This is the moisture measuring and controlling device of the present invention. DETAILED DESCRIPTION

[0033] In order to better explain the present invention, the following Figure 1-4 The technical solution of the present invention is described in detail. The following embodiments are explanations of the invention, and the present invention is not limited to the following embodiments.

[0034] See also Figure 1 The intelligent moisture detection and control method designed by the present invention sets multiple moisture sensors, multiple temperature sensors and multiple humidity sensors in the monitoring environment, and outputs the moisture prediction value after processing by the moisture detection module. The specific moisture detection module is designed as follows:

[0035] The moisture detection module includes GWO's NARX neural network model-GWO's BiLSTM neural network model 1-3, GWO's fuzzy recursive neural network model-GWO's NARX neural network model 1 and GWO's fuzzy recursive neural network model-GWO's BiLSTM neural network model 1.

[0036] 1. GWO's NARX neural network model - GWO's BiLSTM neural network model 1-3 design

[0037] GWO's NARX neural network model - GWO's BiLSTM neural network model is a series connection of GWO's NARX neural network model and GWO's BiLSTM neural network model. GWO's NARX neural network model optimizes GWO's NARX neural network model parameters, and GWO's BiLSTM neural network model optimizes GWO's BiLSTM neural network model parameters.

[0038] GWO optimizes the parameters of the NARX neural network model. GWO is the Grey Wolf Optimization Algorithm. As an intelligent algorithm for global optimization search, the Grey Wolf Optimization Algorithm (GWO) has the advantages of fast convergence speed, few parameters, simple and easy implementation, and has adaptive adjustment factors and information feedback mechanisms. The parameters of the NARX neural network model are to set the number of hidden layers, the number of neurons in each layer, the learning rate, the batch size and other parameters. The GWO optimization algorithm process is as follows:

[0039] (1) Initialize the population, randomly assign values ​​to the population, calculate the fitness of each individual gray wolf, select the three best individuals in the population and record them as α, β and δ, and use them to dominate the other individuals in the population.

[0040] (2) Surround the prey. During the entire optimization process, the population updates its position and surrounds the prey based on the distance between the individual gray wolf and the prey.

[0041] (3) Hunting prey: Under the leadership of the alpha wolves α, β, and δ, other individuals in the population approach the prey, updating the distance between the gray wolf individuals and the prey.

[0042] (4) When the maximum number of iterations is reached or the number of iterations of the gray wolf population exceeds the limit within a certain range, the iteration is stopped and the optimal parameter combination of the NARX neural network model is obtained.

[0043] The NARX neural network model needs the water parameter change sequence values ​​at historical moments as input for the water parameter time series prediction, and also needs the water parameter real-time change sequence as additional input; the NARX neural network model includes three parts: water parameter input layer, water parameter hidden layer and water parameter output layer. The input layer of the NARX neural network model takes into account the water parameter delay lag, increases the dynamic memory capacity of the input water parameter, and the water parameter input reflects the change parameters of the input water parameter at historical moments. By feeding back the water parameter output layer to the water parameter input layer, the dynamic memory capacity of the short-term output water historical data is enhanced. The closed-loop time-delay network model of the NARX neural network model provides the water parameter prediction value as a new response input to the NARX neural network model for dynamic prediction of the water parameter time series, realizing the memory performance of water parameter prediction. The NARX neural network model is a dynamic prediction neural network, and its structure can effectively realize the discrete nonlinear autoregressive system prediction with external input, and improve the dynamic, robust and accurate prediction of water parameter.

[0044] The BiLSTM neural network model of GWO adds moisture parameter back propagation on the basis of its original moisture parameter forward propagation, which improves the time series analysis capability of moisture parameters of the BiLSTM neural network model of GWO and enhances the accuracy of moisture parameter prediction.

[0045] The BiLSTM neural network model of GWO optimizes the parameters of the BiLSTM neural network model for GWO. The BiLSTM neural network model of GWO is mainly composed of an input gate, a forget gate, an output gate, and a unit state. A "gate" is a device that can select whether the moisture parameter or information passes through. Its output range is 0-1. When the value is 0, it means that the moisture parameter cannot pass normally at all, so the moisture parameter or information is intercepted and cannot be transmitted further; when the value is 1, it means that the moisture parameter or information can pass through completely. The function of the input gate is to update the state of the BiLSTM neural network model of GWO, and pass the original aquatic parameter information in the BiLSTM neural network model of GWO and the newly passed moisture parameter information of the BiLSTM neural network model of GWO into the activation function sigmoid and tanh function together, and compress the moisture parameter in the BiLSTM neural network model of GWO. Multiply the two results, and the former determines the importance and retention of the latter moisture parameter. The function of the forget gate is to determine the retention of the moisture parameter information in the BiLSTM neural network model of GWO. The closer the output of the activation function sigmoid is to 1, the more it should be retained, and the closer it is to 0, the more it should be cleared. The cell state mainly plays the role of a pathway, and the moisture parameter information is forward and backward propagated in the BiLSTM neural network model of GWO. The output gate can determine the internal state of the next GWO BiLSTM neural network model, and the model data is fed back to the input through the output gate to form a complete closed loop. Since the propagation path of the moisture parameter is longer, the bidirectional propagation can intercept the forward and reverse moisture parameter characteristics. When the moisture parameter is forwardly transmitted in the model, the model will record the forward transmitted moisture parameter characteristic information, and when the moisture parameter is reversely propagated, the model will record the reverse moisture parameter characteristic information. GWO optimizes the BiLSTM neural network model parameters with reference to the method and steps of GWO optimizing the NARX neural network model parameters of this patent.

[0046] 2. GWO's fuzzy recursive neural network model - GWO's NARX neural network model 1 design

[0047] The fuzzy recursive neural network model of GWO - the NARX neural network model of GWO is a fuzzy recursive neural network model of GWO connected in series with the NARX neural network model of GWO. The fuzzy recursive neural network model of GWO is a GWO optimized fuzzy recursive neural network model parameter. The GWO optimized fuzzy recursive neural network model refers to the method and steps of the GWO optimized NARX neural network model of this patent.

[0048] The fuzzy recursive neural network model is the product of the combination of fuzzy theory and neural network. It has the fuzzy reasoning ability of the moisture parameter fuzzy system and the ability of the neural network to adapt to the changes in moisture parameters. The model can not only automatically update, but also continuously correct the membership function of the fuzzy subset. The recursive neural network increases the ability of the network itself to process the dynamic information of moisture parameters by adding some additional internal feedback channels to the fuzzy recursive neural network model. It can use the internal state of the network to feedback the changes in moisture parameters to describe the nonlinear dynamic behavior of the moisture parameter system, and can more directly reflect the dynamic characteristics of the moisture parameter system, that is, adding self-feedback connections to the rule layer of the forward fuzzy network, and recording the historical information of moisture parameters of the rule activation intensity. Tests show that the fuzzy recursive neural network model is applied to the dynamic modeling of moisture parameters, the test moisture parameter error is small, and the prediction accuracy of moisture parameters is high.

[0049] The fuzzy recursive neural network model includes a moisture parameter input layer, a moisture parameter membership function layer, a moisture parameter rule layer, a moisture parameter recursive layer and a moisture parameter output layer. The moisture parameter input layer: This layer has a total of n neurons, and each node represents a moisture parameter input variable. The purpose is to directly transmit the moisture parameter input value to the next layer. Moisture parameter membership function layer: Each node in this layer represents a moisture parameter membership function. This patent adopts a Gaussian membership function. Moisture parameter rule layer: Each node in this layer represents a moisture parameter fuzzy logic rule, and the fuzzy operator is used as the multiplication operator. Moisture parameter recursive layer: This layer establishes a self-feedback connection to enhance the dynamic ability of the network, introduces internal variables, and selects a linear sum function as the activation function. The number of its nodes is the same as the number of nodes in the moisture parameter rule layer. Moisture parameter output layer: This layer sums its moisture parameter input to achieve defuzzification.

[0050] The design method of GWO's NARX neural network model refers to the relevant design steps of this patent.

[0051] 3. GWO's fuzzy recurrent neural network model - GWO's BiLSTM neural network model 1 design

[0052] The fuzzy recursive neural network model of GWO - the BiLSTM neural network model of GWO is a series connection of the fuzzy recursive neural network model of GWO and the BiLSTM neural network model of GWO. The design method of the fuzzy recursive neural network model of GWO and the BiLSTM neural network model of GWO refers to the relevant design steps of this patent.

[0053] 2. Moisture Control Module Design

[0054] See also Figure 2The moisture control module includes GWO's NARX neural network model-GWO's BiLSTM neural network model 4-5, GWO's fuzzy recursive neural network model-GWO's NARX neural network model 2, GWO's fuzzy recursive neural network model-GWO's BiLSTM neural network model 2-4, a moisture detection module and a PID controller 1-3.

[0055] GWO's NARX neural network model - GWO's BiLSTM neural network model 4-5 has the same design structure as GWO's NARX neural network model - GWO's BiLSTM neural network model 1-3. GWO's fuzzy recursive neural network model - GWO's NARX neural network model 2 has the same design structure as GWO's fuzzy recursive neural network model - GWO's NARX neural network model 1. GWO's fuzzy recursive neural network model - GWO's BiLSTM neural network model 2-4 has the same design structure as GWO's fuzzy recursive neural network model - GWO's BiLSTM neural network model 1.

[0056] The PID controller essentially relies on the past moisture parameter error, the current moisture parameter error, and the prediction of the future moisture parameter error to control the moisture regulation system. The integral part acts on the average of the past moisture parameter error, the proportional part acts on the current moisture parameter error, and the differential part acts on the prediction of the future moisture parameter error based on linear extrapolation. Its input signal e(t) is the difference between the given moisture parameter target value and the moisture parameter feedback value, and the output signal u(t) is the control amount of the moisture parameter. The output of the PID controller is:

[0057]

[0058] Among them, u(t) is the control signal output by the PID controller at time t, that is, the control amount of the moisture parameter, k p is the proportionality coefficient, k i is the integration coefficient, k d is the differential coefficient, e(t) is the error between the controller input signal and the expected value, is the derivative of the error, that is, the rate of change of the error. p The larger the ratio, the more obvious the link is, the faster the water regulation system responds, and the more stable the water regulation system is. p The error of moisture parameter increases with the increase of k i The smaller the k is, the more obvious the effect on the integral link will be, which can continuously reduce the steady-state error of the moisture parameter regulation system. However, it will affect the response speed of the moisture parameter regulation system to a certain extent, thereby prolonging the moisture parameter regulation time. dThe larger the moisture parameter overshoot is, the more obvious the suppression effect is, and the smaller the attenuation effect of the moisture parameter regulation system oscillation is. d Too large a value will slow down the response of the moisture parameter regulation system.

[0059] For PID controller 1, e(t) represents the error between the moisture target value and the output of GWO's NARX neural network model-GWO's BiLSTM neural network model 5. It represents the error change rate.

[0060] For PID controller 2, e(t) represents the error between the water flow rate output by the water pump driving device and the moisture prediction value output by the moisture detection module. It represents the error change rate.

[0061] For PID controller 3, e(t) represents the error between the output of GWO's fuzzy recursive neural network model - GWO's BiLSTM neural network model 3 and the output of flow sensor 2. It represents the error change rate.

[0062] 3. Design of IoT system for moisture detection and control

[0063] The Internet of Things system for moisture detection and control consists of a moisture meter, a moisture cloud platform server and a moisture monitoring platform. The moisture meter collects moisture, temperature, humidity, wind speed and light intensity. The moisture meter transmits the collected data to the moisture cloud platform server through a 4G communication module. The moisture cloud platform server processes and stores the received data and sends the data to the moisture monitoring platform. The moisture monitoring platform dynamically displays, queries, monitors and issues moisture adjustment instructions to the moisture cloud platform server. The moisture cloud platform server transmits the data to the moisture meter. The moisture monitoring platform includes a moisture monitoring terminal and a moisture monitoring APP terminal. The structure of the Internet of Things system for moisture detection and control is shown in Figure 3 shown.

[0064] 1. Moisture measurement and control device design

[0065] The moisture measurement and control device includes a temperature sensor, an illumination sensor, a wind speed sensor, a moisture sensor and their corresponding conditioning circuits, a visual sensor, a water pump drive device, an electric valve and a 4G communication module. The moisture measurement and control device based on the 4G communication module realizes information interaction with the moisture cloud platform server. The software of the moisture measurement and control device mainly realizes the moisture detection module, the moisture control module, the 4G communication module and the collection and preprocessing of sensor parameters and the water supply control module. Figure 4 .

[0066] 2. Moisture monitoring platform design

[0067] The moisture monitoring platform includes a moisture monitoring terminal and a moisture monitoring APP terminal. The moisture monitoring terminal is an industrial control computer. The moisture monitoring terminal realizes information exchange with the moisture cloud platform server through the 5G network. It mainly realizes the collection, processing and monitoring of environmental parameters, moisture, water pump drive device and electric valve detection and control parameters. The main functions of the moisture monitoring terminal are communication parameter setting, environmental parameter management and moisture monitoring. The moisture monitoring APP terminal realizes information exchange with the moisture cloud platform server through the 5G network. The moisture monitoring APP terminal mainly realizes the collection and monitoring of the environment, water quality, water pump drive device and electric valve detection and control parameters.

[0068] The above embodiments are only for illustrating the technical concept and features of the present invention, and their purpose is to enable people familiar with the technology to understand the content of the present invention and implement it accordingly, and they cannot be used to limit the protection scope of the present invention. Any equivalent transformation or modification made according to the spirit of the present invention should be included in the protection scope of the present invention.

Claims

1. An intelligent moisture detection and control method, characterized in that: The steps include: Step 1: Set up multiple moisture sensors, multiple temperature sensors and multiple humidity sensors in the monitoring environment; Step 2: Construct a moisture detection module, process the collected values ​​of multiple moisture sensors, multiple temperature sensors and multiple humidity sensors through the moisture detection module and output the moisture prediction value; The moisture detection module includes GWO's NARX neural network model-GWO's BiLSTM neural network model 1-3, GWO's fuzzy recursive neural network model-GWO's NARX neural network model 1 and GWO's fuzzy recursive neural network model-GWO's BiLSTM neural network model 1; The outputs of multiple moisture sensors, multiple temperature sensors and multiple humidity sensors are respectively used as inputs of GWO's NARX neural network model-GWO's BiLSTM neural network model 1-3, the outputs of multiple humidity sensors and the outputs of GWO's NARX neural network model-GWO's BiLSTM neural network model 1-3 are used as corresponding inputs of GWO's fuzzy recursive neural network model-GWO's NARX neural network model 1, the outputs of multiple moisture sensors, the outputs of GWO's NARX neural network model-GWO's BiLSTM neural network model 1 and the outputs of GWO's fuzzy recursive neural network model-GWO's NARX neural network model 1 are used as corresponding inputs of GWO's fuzzy recursive neural network model-GWO's BiLSTM neural network model 1, and the outputs of GWO's fuzzy recursive neural network model-GWO's BiLSTM neural network model 1 are used as the moisture prediction value of the detected object; Step 3: Set up a water pump driving device, an electric valve and a flow sensor 1-2, and the flow sensor 1-2 measures the flow rate of water driven and output by the water pump driving device and the flow rate of water controlled and output by the electric valve respectively; Step 4: According to the moisture prediction value and the moisture target value, the moisture control module controls the control instructions of the water pump drive device and the electric valve, and adjusts the moisture value of the detected object in real time to make it meet the moisture target value.

2. An intelligent moisture detection and control method according to claim 1, characterized in that: The moisture control module includes a NARX neural network model of GWO-a BiLSTM neural network model of GWO 4-5, a fuzzy recursive neural network model of GWO-a NARX neural network model of GWO 2, a fuzzy recursive neural network model of GWO-a BiLSTM neural network model of GWO 2-4, the moisture detection module and a PID controller 1-3; The moisture target value is used as the input of GWO's NARX neural network model - GWO's BiLSTM neural network model 4, the error and error change rate between the moisture target value and the output of GWO's NARX neural network model - GWO's BiLSTM neural network model 5 are used as the input of GWO's fuzzy recurrent neural network model - GWO's NARX neural network model 2 and the input of PID controller 1, and the output of PID controller 1 and the output of GWO's NARX neural network model - GWO's BiLSTM neural network model 4 are used as GWO's fuzzy recurrent neural network model - GWO's BiLSTM neural network model 2 The output of GWO's fuzzy recursive neural network model - GWO's BiLSTM neural network model 2 is used as the input of the water pump drive device and the input of GWO's fuzzy recursive neural network model - GWO's BiLSTM neural network model 3. The error and error change rate of the water flow output by the water pump drive device and the moisture prediction value output by the moisture detection module are used as the input of PID controller 2 and GWO's fuzzy recursive neural network model - GWO's NARX neural network model 2. The output of PID controller 2 is used as the input of GWO's fuzzy recursive neural network model - GWO's BiLSTM neural network model 3. The output of BiLSTM neural network model 3 of WO-GWO is used as the input of BiLSTM neural network model 4 of GWO. The error and error change rate between the output of BiLSTM neural network model 3 of GWO-GWO and the output of flow sensor 2 are used as the input of NARX neural network model 2 of WO-GWO and the input of PID controller 3. The output of PID controller 3 is used as the input of BiLSTM neural network model 4 of GWO-GWO. The output of WO's fuzzy recursive neural network model - GWO's BiLSTM neural network model 4 is used as the input of GWO's fuzzy recursive neural network model - GWO's BiLSTM neural network model 2-3 and the input of the electric valve. The electric valve outputs the water flow to intelligently adjust the moisture of the moisture regulation object. The output of the moisture sensor group is used as the input of the moisture detection module. The output of the flow sensor 1-2 and the output of the moisture detection module are used as the input of GWO's NARX neural network model - GWO's BiLSTM neural network model 5. GWO's fuzzy recursive neural network model - GWO's NARX neural network model 2 outputs 3 groups of k p , k i and k d As k of PID controller 1-3 respectively p , k i and k d ; The moisture sensor group includes a plurality of moisture sensors, a plurality of temperature sensors and a plurality of humidity sensors arranged in the monitoring environment.

3. An intelligent moisture detection and control method according to claim 2, characterized in that: The outputs of the PID controllers 1-3 are: Among them, u(t) is the control signal output by the PID controller at time t, that is, the control amount of the moisture parameter, k p is the proportionality coefficient, k i is the integration coefficient, k d is the differential coefficient, e(t) is the error between the controller input signal and the expected value, is the derivative of the error, that is, the rate of change of the error; k p The larger the ratio, the more obvious the link, the faster the water regulation response, and the more stable the water regulation. p The increase of moisture parameter error increases; k i The smaller the k is, the more obvious the effect on the integral link is, which makes the steady-state error of moisture parameter adjustment continuously reduce, affects the response speed of moisture parameter adjustment to a certain extent, and prolongs the moisture parameter adjustment time; d The larger the moisture parameter overshoot is, the more obvious the suppression effect is, making the attenuation effect of moisture parameter regulation oscillation smaller. d Too large a value will slow down the response speed of moisture parameter adjustment.

4. The intelligent moisture detection and control method according to claim 2 is characterized in that: The GWO NARX neural network model - GWO BiLSTM neural network model 1-5 are all GWO NARX neural network model and GWO BiLSTM neural network model connected in series. The GWO NARX neural network model is for GWO to optimize the NARX neural network model parameters, and the GWO BiLSTM neural network model is for GWO to optimize the BiLSTM neural network model parameters.

5. The intelligent moisture detection and control method according to claim 2, characterized in that: The GWO fuzzy recursive neural network model-GWO NARX neural network model 1-2 are both GWO fuzzy recursive neural network models and GWO NARX neural network models connected in series, the GWO fuzzy recursive neural network model is GWO optimized fuzzy recursive neural network model parameters, and the GWO NARX neural network model is GWO optimized NARX neural network model parameters; The fuzzy recursive neural network model of GWO - the BiLSTM neural network model of GWO 1-4 are all fuzzy recursive neural network models of GWO and BiLSTM neural network models of GWO connected in series. The fuzzy recursive neural network model of GWO is the GWO optimized fuzzy recursive neural network model parameters, and the BiLSTM neural network model of GWO is the GWO optimized BiLSTM neural network model parameters; the fuzzy recursive neural network model is a model combining fuzzy theory and neural network.

6. An Internet of Things system for moisture detection and control, characterized in that: It includes a moisture measurement and control device, a moisture cloud platform server and a moisture monitoring platform; the moisture measurement and control device collects moisture, temperature, humidity, wind speed and illumination, and transmits the collected data to the moisture cloud platform server through a 4G communication module; the moisture cloud platform server processes and stores the received data, and sends the data to the moisture monitoring platform at the same time; the moisture monitoring platform dynamically displays, queries, monitors and issues moisture adjustment instructions to the moisture cloud platform server for the data of the moisture measurement and control device, and the moisture cloud platform server transmits the instructions to the moisture measurement and control device; the moisture cloud platform server executes the process of the intelligent moisture detection and control method as described in any one of claims 1 to 5 when processing and storing the received data.

7. The Internet of Things system for moisture detection and control according to claim 6, characterized in that: The moisture monitoring platform includes a moisture monitoring terminal and a moisture monitoring APP terminal; the moisture monitoring terminal is an industrial control computer, and the moisture monitoring terminal and the moisture monitoring APP terminal both dynamically display, query, monitor and issue moisture adjustment instructions to the moisture cloud platform server for the data of the moisture measuring and controlling device.

8. An Internet of Things device for moisture detection and control, comprising a memory and a processor, wherein a computer program is stored in the memory, characterized in that: When the computer program is executed by a processor, the process of the intelligent moisture detection and control method according to any one of claims 1 to 5 is implemented.

Images