A greenhouse environment remote monitoring method applied to greenhouse planting

Abstract

The application discloses a greenhouse environment remote monitoring method applied to greenhouse planting, and relates to the technical field of big data, and comprises the following steps: a sensor network is built to collect environment parameters; environment data is preprocessed; multi-dimensional environment analysis is conducted according to the cleaned data; environment modeling is conducted according to the analysis result; temperature and humidity prediction and adjustment suggestions are conducted on the environment model; automatic control is conducted according to the adjustment suggestions; system optimization is conducted on the automatic control result; and environment parameter collection results are fed back in real time. The application proposes an improved multilayer perceptron (MLP) network structure, multi-dimensional input and a weighting mechanism are introduced, so that the model can more accurately capture the complex nonlinear relationship between the greenhouse environment parameters. Especially when facing the rapid changes of environmental factors such as temperature and humidity, the improved MLP network can flexibly adjust the weight of the model, and improve the accuracy of the temperature and humidity change trend prediction.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of big data, and particularly to a greenhouse environment remote monitoring method applied to greenhouse planting. BACKGROUND

[0002] With the continuous development of modern agricultural technology, especially in the field of greenhouse planting, the refinement and intelligentization of environmental control technology have become the key to improving crop growth efficiency and quality. The monitoring and regulation system of greenhouse planting environment usually optimizes the environmental parameters such as temperature and humidity, air flow, light, carbon dioxide in the greenhouse by integrating sensors, data acquisition, real-time monitoring, environmental regulation and other technologies.

[0003] In order to realize the intelligentization of environmental regulation, researchers have gradually applied machine learning algorithms such as multilayer perceptron (MLP) and long short-term memory network (LSTM) for data modeling and prediction in recent years. However, the existing technology still faces some challenges, especially in the prediction of temperature and humidity change trend and the modeling of complex relationship between environmental factors.

[0004] In the prior art, the deficiency of multilayer perceptron (MLP) network structure: most of the current MLP-based models applied in greenhouse environment monitoring and regulation usually adopt standard feedforward neural network structure, which directly maps input and output from sensor data. Although this structure can capture the basic relationship between environmental factors and crop growth, it has limitations in dealing with highly nonlinear, complex time series changes and interactions between environmental factors. The existing MLP model cannot fully explore the time series dependence relationship between multi-dimensional environmental parameters, resulting in insufficient prediction accuracy, especially in the case of dramatic changes in environmental parameters, the robustness and adaptability of the model are poor. SUMMARY

[0005] In view of the deficiencies of the prior art, the present application provides a greenhouse environment remote monitoring method applied to greenhouse planting to solve the problems raised in the background art.

[0006] To achieve the above purpose, the present application provides the following technical scheme:

[0007] In the first aspect, the present application provides a greenhouse environment remote monitoring method applied to greenhouse planting, comprising the following steps:

[0008] S1, a sensor network is built to collect environmental parameters;

[0009] A sensor network based on Internet of Things is established to collect environmental parameters in the greenhouse to obtain environmental data;

[0010] S2, pre-processing the environmental data;

[0011] Correcting the environmental data through data preprocessing techniques to obtain cleaned data;

[0012] S3, performing multi-dimensional environmental analysis according to the cleaned data;

[0013] Using principal component analysis techniques to identify the relationship between the environmental data in the greenhouse, and obtaining the analysis results;

[0014] S4, performing environmental modeling according to the analysis results;

[0015] Based on the multi-layer perceptron (MLP) network structure model, a prediction model between the environmental data in the greenhouse and the crop growth is constructed, and an environmental model is obtained;

[0016] S5, temperature and humidity prediction and adjustment suggestions for the environmental model;

[0017] Based on the long short-term memory (LSTM) network, the temperature and humidity of the environmental model are predicted and adjustment suggestions are generated, and the adjustment suggestions are obtained;

[0018] S6, performing automatic control according to the adjustment suggestions;

[0019] Using an automatic control system based on the industrial automation communication protocol Modbus protocol to realize remote control of the temperature and humidity adjustment equipment, and obtaining the automatic control results;

[0020] S7, based on the automatic control results, collecting real-time environmental parameters and feeding back the collection results in real time.

[0021] Further optimize the technical solution, in step S1, the environmental parameters include:

[0022] Temperature and humidity, air flow speed, light intensity, and carbon dioxide concentration.

[0023] Further optimize the technical solution, in step S2, the data preprocessing techniques include:

[0024] Data interpolation: using the mean, median or mode of the environmental data to fill in the missing data;

[0025] Outlier detection: calculating the standard score of the environmental data points, and if the difference between the environmental data points and the mean is greater than twice or three times the standard deviation, the point is considered to be an outlier;

[0026] Data smoothing: a filtering technique to reduce noise, which calculates the average value within a window around the environmental data points to reduce noise.

[0027] Further optimize the technical solution, in step S3, the principal component analysis techniques include:

[0028] The direction with the maximum variance in the cleaning data is searched to extract the main features, and the environmental parameters with the maximum variance in the cleaning data have the greatest influence on crop growth.

[0029] Further optimization of the technical solution, the step S4, the multi-layer perceptron MLP network structure model comprises:

[0030] Establishing an environmental model of environmental parameters in the greenhouse on crop growth, introducing a time weighting mechanism, temperature and humidity prediction and analysis.

[0031] Further optimization of the technical solution, the step S4, the multi-layer perceptron MLP network structure model comprises:

[0032] ;

[0033] Wherein,

[0034] is the crop growth state output at time ;

[0035] is the input environmental parameter, including temperature , humidity , air flow velocity , light intensity and carbon dioxide concentration at time ;

[0036] is a preset value, representing the time-varying weight coefficient of each input parameter, and the relative influence of each environmental factor on crop growth at different time points;

[0037] is a bias term, representing the basic output value of the network, and the initial value is 0;

[0038] is a nonlinear activation function, and its function expression is .

[0039] Further optimization of the technical solution, the step S4, the multi-layer perceptron MLP network structure model comprises:

[0040] The time window parameter is introduced, so that each input depends not only on the data at the current time, but also on the environmental data in the history time, which is represented by the following formula:

[0041] ;

[0042] Wherein,

[0043] is the length of the time window, the crop growth state at the current time depends not only on the environmental data at the current time, but also on the environmental data at the past time;

[0044] represents the weighted sum of the environmental parameters at the past time;

[0045] is the crop growth state output at the past time.

[0046] Further optimization of the present technical solution, the temperature and humidity prediction and analysis includes:

[0047] Data collection: through the cleaning data and analysis results, and record according to time sequence;

[0048] Model training: input the collected data, using multilayer perceptron MLP network structure model training, learning the nonlinear relationship between the input environmental factors and temperature and humidity change;

[0049] Temperature and humidity prediction: after training, using the new real-time environmental data including the current time of temperature and humidity, air flow speed, light intensity, get the future time of temperature and humidity prediction result;

[0050] Environmental regulation suggestion: according to the predicted temperature and humidity prediction result, adjust the temperature and humidity control equipment in the greenhouse, in order to maintain the suitable crop growth environment.

[0051] Further optimization of the present technical solution, in step S5, the long short-term memory network LSTM includes:

[0052] Using multilayer perceptron MLP network structure model to get the environmental prediction model in the greenhouse;

[0053] Multi-dimensional joint processing is adopted, all environmental factors are jointly modeled and the environmental prediction model is used, and the activation function is used for each environmental factor, which is the memory unit of long short-term memory network LSTM.

[0054] The time window weighting mechanism is adopted, that is, the historical data in each time window is weighted according to the influence degree of environmental factors. The influence of the environmental factor on the crop growth environment is determined by calculating the variance, the greater the variance, the greater the influence.

[0055] ​Further optimization of the technical solution, in step S7, the sensor network established by step S1 collects the environmental parameters in the greenhouse in real time, compares the difference between the results of the device adjustment after the automatic control and the actual environmental parameters, and provides a reference basis for the adjustment of the device.

[0056] In a second aspect, the embodiments of the present application provide a computer device, comprising a memory and a processor, the memory stores a computer program, wherein: the computer program instructions are executed by the processor to realize the steps of the greenhouse environment remote monitoring method applied to greenhouse planting according to the first aspect of the present application.

[0057] In a third aspect, the embodiments of the present application provide a computer readable storage medium, which stores a computer program, wherein: the computer program instructions are executed by the processor to realize the steps of the greenhouse environment remote monitoring method applied to greenhouse planting according to the first aspect of the present application.

[0058] Compared with the prior art, the present application provides a greenhouse environment remote monitoring method applied to greenhouse planting, which has the following beneficial effects:

[0059] The greenhouse environment remote monitoring method applied to greenhouse planting proposes an improved multi-layer perceptron (MLP) network structure, which can more accurately capture the complex nonlinear relationship between greenhouse environmental parameters (such as temperature and humidity, air flow speed, light intensity, carbon dioxide concentration, etc.) by introducing multi-dimensional input and weighting mechanism. In this improved structure, by jointly modeling multi-dimensional environmental factors, the network can more comprehensively understand the interaction effects between different environmental factors. Especially when facing rapid changes in environmental factors such as temperature and humidity, the improved MLP network can flexibly adjust the weight of the model, improving the accuracy of temperature and humidity trend prediction. This not only enhances the prediction accuracy of the model, but also improves the real-time and stability of greenhouse environment regulation, effectively optimizing the crop growth environment. BRIEF DESCRIPTION OF DRAWINGS

[0060] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced as follows. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0061] Fig. 1 A flowchart of a greenhouse environment remote monitoring method applied to greenhouse planting is provided for the present application;

[0062] Fig. 2A temperature and humidity prediction and analysis flowchart of a greenhouse environment remote monitoring method applied to greenhouse planting is provided for the present application.

[0063] Fig. 3 A multi-layer perceptron MLP network structure model formula model usage and explanation diagram of a greenhouse environment remote monitoring method applied to greenhouse planting is provided for the present application. DETAILED DESCRIPTION

[0064] In order to make the above-mentioned objects, features and advantages of the present application more apparent and easy to understand, the specific embodiments of the present application will be described in detail below with reference to the accompanying drawings.

[0065] In the following description, many specific details are set forth in order to provide a thorough understanding of the present application. However, the present application can be practiced in other ways that are not the same as the description, and those skilled in the art can make similar extensions without departing from the spirit of the present application, so the present application is not limited to the specific embodiments disclosed below.

[0066] Secondly, the "one embodiment" or "embodiment" referred to herein means that the specific features, structures or characteristics can be included in at least one implementation of the present application. "In one embodiment" appearing in different places in the specification does not mean the same embodiment, nor is it an embodiment that is separate or selectively excluded from other embodiments.

[0067] Embodiment one:

[0068] Reference Figs. 1-3 For the first embodiment of the present application, the embodiment provides a greenhouse environment remote monitoring method applied to greenhouse planting, comprising the following steps:

[0069] S1, build a sensor network to collect environmental parameters;

[0070] A sensor network based on Internet of Things is established to collect environmental parameters in the greenhouse to obtain environmental data.

[0071] The environmental parameters include:

[0072] Temperature and humidity, air flow speed, light intensity and carbon dioxide concentration.

[0073] In this embodiment, the sensor data acquisition module transmits various environmental data in the greenhouse to the central processing unit in real time. These data are crucial for subsequent environmental analysis and model construction. The technology used is LoRaWAN network technology, which can transmit data over a large range and has the advantages of low power consumption and long distance transmission, especially suitable for long-term operation of devices in the greenhouse environment.

[0074] S2, preprocessing the environmental data;

[0075] The environmental data is corrected by data preprocessing techniques to obtain cleaned data. The data preprocessing techniques include:

[0076] Data interpolation: using the mean, median or mode of the environmental data to fill in missing values;

[0077] Outlier detection: calculate the standard score of the environmental data points, if the difference between the environmental data point and the mean is greater than twice or three times the standard deviation, then the point is considered as an outlier;

[0078] Data smoothing: a filtering technique to reduce the impact of noise by calculating the average value within a window around the environmental data point.

[0079] In this embodiment, the environmental data collected by the sensor will face the problems of noise, missing values and outliers, so it is necessary to preprocess and clean the data.

[0080] S3, multi-dimensional environmental analysis based on cleaned data;

[0081] Using principal component analysis technique to identify the relationship between the environmental data in the greenhouse, and obtaining the analysis result.

[0082] The principal component analysis technique includes:

[0083] Find the direction with the largest variance in the cleaned data to extract the main features, the environmental parameters with the largest variance in the cleaned data have the greatest impact on crop growth.

[0084] In this embodiment, by analyzing the correlation between temperature, humidity, air flow speed and crop growth, it can be found out which environmental factors have the greatest impact on crop growth. Using principal component analysis (PCA) technique, the multi-dimensional data is reduced, the main influencing factors are found out, and the input data for the next step of intelligent model is provided. Through PCA, the calculation amount can be effectively reduced, and the accuracy of the analysis result can be ensured.

[0085] S4, environmental modeling based on analysis results;

[0086] Based on the multi-layer perceptron (MLP) network structure model, a prediction model between the environmental data in the greenhouse and the crop growth is constructed, and an environmental model is obtained.

[0087] The multi-layer perceptron (MLP) network structure model includes:

[0088] Establishing an environmental model of the environmental parameters in the greenhouse on crop growth:

[0089] ;

[0090] Among them,

[0091] is the crop growth state output at time ;

[0092] is the input environmental parameters including temperature , humidity , air flow speed , light intensity and carbon dioxide concentration at time ;

[0093] is the preset value, representing the time-varying weight coefficient of each input parameter, the relative influence of each environmental factor on crop growth at different time points;

[0094] is the bias term, representing the basic output value of the network, the initial value is 0;

[0095] is the nonlinear activation function, and its function expression is .

[0096] The time sequence weighting mechanism is introduced,

[0097] so that each input not only depends on the data at the current time, but also depends on the environmental data in the history time.

[0098] It is expressed by the following formula:

[0099] ;

[0100] wherein,

[0101] is the length of the time window, indicating that the crop growth state at the current time not only depends on the environmental data at the current time, but also depends on the environmental data at the past time;

[0102] represents the weighted sum of the environmental parameters in the past time;

[0103] is the crop growth state output in the past time.

[0104] Temperature and humidity prediction and analysis.

[0105] Data collection: through cleaning data and analysis results, and recording in time sequence;

[0106] Model training: Input the collected data, use the multi-layer perceptron (MLP) network structure model for training, learn the nonlinear relationship between input environmental factors and temperature and humidity changes.

[0107] Temperature and humidity prediction: After the model training is completed, the future temperature and humidity are predicted based on the new real-time environmental data. The new real-time environment includes the current temperature and humidity, air flow speed, and light intensity.

[0108] Environmental adjustment suggestion: According to the predicted temperature and humidity trend, adjust the temperature and humidity control equipment in the greenhouse to maintain a suitable crop growth environment.

[0109] In this embodiment, the MLP network structure model can establish a nonlinear relationship between environmental parameters (temperature and humidity, air flow, light intensity, and carbon dioxide concentration) and crop growth status (such as growth stage, leaf area, etc.). Unlike traditional linear regression models, the MLP network structure model can capture complex nonlinear relationships, thus more accurately reflecting the impact of environmental changes on crop growth. This model provides data support for subsequent decision support systems and can predict crop growth changes when environmental parameters change.

[0110] The use of the multi-layer perceptron (MLP) network structure model is as follows:

[0111] Input setting: The input of this model includes four main environmental factors: temperature and humidity, air flow speed, light intensity, and carbon dioxide concentration. Each factor is collected at each time point to form a time series data set. These data are standardized and input into the model.

[0112] Dynamic weighting mechanism: In traditional MLP network structure models, the input weight is fixed, while in this model, the weight is time-varying, meaning that the influence of each environmental factor on crop growth varies at different times. Through historical data learning, these weights can be dynamically adjusted to better reflect the real-time impact of the environment on crop growth.

[0113] Time window mechanism: Considering the long-term impact of environmental changes on crop growth, a time window is introduced into the model, so that the environmental data at each time point is not only related to the current crop growth status, but also to the environmental data in the past period, capturing the influence of long-term trends and short-term fluctuations.

[0114] Output and decision support: The output value (crop growth status) from the network can provide data support for the environmental control system in the greenhouse, adjusting environmental parameters based on the predicted crop growth status. For example, if the model predicts poor crop growth (i.e., Yt is low), the system can automatically adjust the temperature and humidity, light, and other environmental factors to optimize crop growth conditions.

[0115] If you want to use this model for temperature and humidity trend prediction, you can extend and apply it in the following ways:

[0116] Only need to use environmental factors input (such as temperature and humidity, air flow speed, etc.), and modify the output target to the future predicted value of temperature and humidity. The following two strategies can be taken:

[0117] Separate prediction of temperature and humidity changes: predict future temperature and humidity trends by inputting environmental factors (such as air flow speed, light intensity, carbon dioxide concentration) into the model. The output of the model can be directly the predicted temperature and humidity values.

[0118] Joint prediction of crop growth and temperature and humidity: if you need to consider the impact of crop growth and temperature and humidity changes on environmental control, you can design a multi-task learning model based on the model of the impact of greenhouse environmental parameters on crop growth, and add the prediction target of temperature and humidity changes, so that the model can output the crop growth status and environmental change trend at the same time.

[0119] Specific methods and implementation:

[0120] Input adjustment: environmental parameter input: in order to predict temperature and humidity trends, the input will mainly include historical data of temperature and humidity (observed values of temperature and humidity at the past few time points) and other related environmental parameters (such as air flow speed, light intensity, carbon dioxide concentration, etc.), forming a time series data set. Crop growth status: in the prediction process, the crop growth status can be used as an auxiliary input to help understand the impact of the environment on the crop, and then reflect the long-term trend of environmental change.

[0121] When using the environmental model to predict temperature and humidity trends, the model formula needs to be adjusted and applied as follows:

[0122] According to the content in step S4 above, the environmental model of greenhouse environmental parameters on crop growth is

[0123] ;

[0124] On this basis,

[0125] Target output adjustment: set the output of the model Directly set to the predicted value of temperature and humidity, that is, the predicted temperature and humidity at the future time point. For example, if the temperature is predicted, then:

[0126] ;

[0127] Where, is the predicted temperature at the next time instant. Similarly, the humidity prediction can be expressed similarly as:

[0128] ;

[0129] where, is the predicted humidity at the next time instant.

[0130] Introducing time series characteristics: Since temperature and humidity have time series properties, we can introduce a historical data window (e.g., data from the past K time instants) as input to better capture the trend of temperature and humidity changes. For example, the prediction formula for temperature can be extended as:

[0131] ;

[0132] where, represents the temperature and humidity, air flow speed, etc. data at the past time instants.

[0133] Model output and temperature and humidity change trend: After training, the model will be able to predict the temperature and humidity values at future time instants based on the current and past environmental factors (temperature and humidity, air flow speed, light intensity, carbon dioxide concentration, etc.). This output can provide prediction information for the greenhouse environmental control system, guiding the system to adjust the environmental conditions at a certain time in the future to maintain the optimal growth state of crops.

[0134] Traditional temperature and humidity prediction methods usually rely on linear regression or statistical models, which are difficult to handle complex environmental changes and nonlinear relationships. The present application uses an MLP network structure model, which can better model the nonlinear relationship between environmental factors and temperature and humidity, thereby improving the accuracy of prediction. By introducing environmental data over a period of time into the model and considering the time series changes of the environment, the model can capture the trend and periodic characteristics of temperature and humidity changes, thereby improving the reliability of the prediction. By considering multiple environmental factors (such as temperature and humidity, air flow speed, etc.) and using a weighted mechanism to handle the importance of each factor, the model can be more flexible and accurate, adapting to the complexity of the greenhouse environment.

[0135] S5, temperature and humidity prediction and adjustment suggestion for the environment model;

[0136] Based on the long short-term memory network LSTM, the temperature and humidity of the environment model are predicted and adjustment suggestions are generated, obtaining the adjustment suggestions.

[0137] The long short-term memory network LSTM includes:

[0138] The multi-layer perceptron MLP network structure model is used to obtain the environment prediction model in the greenhouse.

[0139] A multi-dimensional joint processing is adopted to jointly model all environmental factors, and activation functions are used for each environmental factor to achieve this, which are the memory units of long short-term memory networks (LSTM). It can model the time-dependent relationship of multi-dimensional data. This joint modeling method can capture the interaction between different environmental factors, such as the joint effect of temperature and humidity and light intensity, the relationship between air flow speed and carbon dioxide concentration, etc., thereby improving the accuracy of prediction.

[0140] A time window weighting mechanism is adopted, that is, the historical data in each time window is weighted according to the influence degree of the environmental factor, and the influence of the environmental factor on the crop growth environment is determined by calculating the variance, the greater the variance, the greater the influence. For example, when the temperature and humidity fluctuate greatly, the model will automatically increase the weight of the temperature and humidity input, so that its influence on the prediction result is more significant; on the contrary, if the air flow speed changes little, its influence on the prediction result will be relatively weak. This mechanism enables the model to adaptively adjust the input weight at each time, improving the robustness of the model in a variable environment.

[0141] In this embodiment, the model formula is calculated according to the following steps:

[0142] Input setting: collect and preprocess the historical data of temperature and humidity, air flow speed, light intensity and carbon dioxide concentration in the greenhouse. Assuming that the time series length of each environmental factor is K, then the input of each environmental factor at time t is a vector of length K, containing the data of the past K time points. For example, the input of temperature and humidity is The air flow speed input is and so on.

[0143] LSTM network training: use the historical data of environmental factors in the greenhouse to train the LSTM model, and update the weights and bias items in the network through the back propagation algorithm (update the network parameters by calculating the gradient of the loss function with respect to these parameters, the purpose is to minimize the loss function). The network automatically learns the nonlinear relationship between different environmental factors and dynamically adjusts the weight of the input data through the time window weighting mechanism, so as to accurately predict the change of future temperature and humidity and other environmental variables.

[0144] Prediction and adjustment suggestion: after training is completed, real-time environmental data is input into the trained LSTM network, and the model will output the prediction results of temperature and humidity, air flow speed or light intensity in the future period. According to the prediction results, the model can generate corresponding adjustment suggestions, such as automatically adjusting the temperature and humidity control equipment, increasing or reducing light, etc., to maintain the best crop growth environment.

[0145] S6, automatically control according to the adjustment suggestion;

[0146] An automatic control system based on the Modbus protocol is used to remotely control the temperature and humidity adjustment equipment, obtaining an automatic control result.

[0147] In this embodiment, the temperature and humidity adjustment equipment comes with an automatic control system when it leaves the factory. The Modbus protocol in this system is a common industrial automation communication protocol that enables interconnection and intercommunication between devices. The temperature and humidity adjustment equipment (such as humidifiers, dehumidifiers, and fans) exchanges real-time data with the control system through the Modbus protocol and automatically performs adjustment operations according to the adjustment suggestion. When the temperature and humidity adjustment equipment is a humidifier, the automatic control system is used to increase humidity; when the temperature and humidity adjustment equipment is a dehumidifier, the automatic control system is used to reduce humidity; when the temperature and humidity adjustment equipment is a fan, the automatic control system is used to adjust the air flow speed. This step ensures real-time adjustment of the greenhouse environment, reduces manual intervention, and improves the efficiency of environmental management.

[0148] S7, real-time feedback based on environmental parameter collection results;

[0149] In step S7, the temperature and humidity adjustment equipment is remotely controlled through the Modbus protocol, but real-time control may result in control errors. In step S7, the sensor network established in step S1 is used to collect real-time environmental information in the greenhouse again, compare the difference between the expected result of the equipment adjustment and the actual environmental information, and provide a reference basis for accurate adjustment of the equipment.

[0150] The process of adjusting environmental parameters and correcting control algorithms includes the following steps:

[0151] Real-time feedback data collection: The sensor network continuously monitors environmental parameters (such as temperature, humidity, and air flow speed) in the greenhouse and compares them with the preset target parameters. When the system detects a deviation between the actual environmental parameters and the target values, it automatically triggers the feedback mechanism.

[0152] Control error analysis and adjustment: Based on the feedback data, the deviation between the actual output of the temperature and humidity control equipment and the expected effect is analyzed. By comparing the real-time monitoring data with the preset target, the error value is calculated, and the control amount that needs to be corrected is generated. This error value is fed back to the adjustment equipment, indicating that it needs to be fine-tuned.

[0153] Dynamic adjustment of equipment operation parameters: based on the environmental change prediction results generated by the multi-layer perceptron (MLP) and long short-term memory network (LSTM) in steps S4 and S5, combined with real-time feedback data, dynamically adjust the operation mode of the control equipment. For example, if the temperature and humidity adjusting equipment has a slow response speed, the system can adjust the opening duration and frequency of the equipment, or change the operation mode of the heating / humidifying equipment, to more accurately control the environmental parameters;

[0154] Collect real-time feedback data through the sensor network to obtain the actual environmental parameters and the preset target value . The error is defined as the difference between the two:

[0155] ;

[0156] wherein, represents the error value at time , and is the real-time monitored environmental parameter, which can be any one of temperature , humidity , air flow speed , light intensity and carbon dioxide concentration , and is the target parameter preset by the system;

[0157] In order to correct the error, the control system adjusts the working parameters of the equipment according to the calculated error. The control amount and the error can be modeled by the following correction formula:

[0158] ;

[0159] wherein, is the feedback gain, is the weighted factor of the control amount at the previous time, is the control output at the previous time. This model reflects that the system combines the current error and historical control results when adjusting the control amount, to more smoothly and effectively adjust the equipment.

[0160] Based on the environmental change prediction results generated by the multi-layer perceptron (MLP) and long short-term memory network (LSTM) model in steps S4 and S5, predict the future temperature and humidity change trend. Assuming that the future environmental state prediction obtained by the LSTM is , then the corrected control output can be further adjusted according to the prediction result:

[0161] ;

[0162] wherein, is a nonlinear function that fuses the error and the prediction result, dynamically adjusts the device operating parameters according to the changes in the environment, and ensures the stability and accuracy of the system.

[0163] Model usage and application:

[0164] Real-time error calculation: by calculating , the system can detect environmental changes in real time and generate feedback.

[0165] Control correction: according to the formula , adjust the working strength of the control device (such as heater, humidity regulator, etc.) according to the feedback error.

[0166] Prediction combination: combine the environmental trend predicted by the LSTM model to dynamically correct the control strategy, avoiding the problem of inaccurate control when the environment changes dramatically.

[0167] Example two:

[0168] The embodiment also provides a computer device suitable for the greenhouse environment remote monitoring method applied to greenhouse planting, which comprises a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute the computer executable instructions to realize the greenhouse environment remote monitoring method applied to greenhouse planting as proposed in the above embodiment.

[0169] The embodiment also provides a storage medium having a computer program stored thereon, which is executed by a processor to realize the greenhouse environment remote monitoring method applied to greenhouse planting as proposed in the above embodiment.

[0170] The computer device can be a terminal, which comprises a processor, a memory, a communication interface, a display screen and an input device connected through a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device comprises a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with external terminals in a wired or wireless manner. The wireless manner can be achieved through WIFI, operator network, NFC (near field communication) or other technologies. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer overlaid on the display screen, or a key, trackball or touchpad arranged on the shell of the computer device, or an external keyboard, touchpad or mouse, etc.

[0171] If the functions are implemented in software, the functions can be stored in or implemented as one or more instructions or code on a computer-readable medium. Computer-readable media include both computer storage media and communication media including any medium that facilitates transfer of a computer program from one place to another. A storage medium can be any available medium that can be accessed by a computer. By way of example, and not limitation, such computer-readable media can comprise RAM, ROM, EEPROM, CD-ROM or other optical disk storage, magnetic disk storage or other magnetic storage devices, or any other medium that can be used to carry or store desired program code in the form of instructions or data structures and that can be accessed by a computer. Also, any connection is properly termed a computer-readable medium. For example, if the software is transmitted from a website, server, or other remote source using a coaxial cable, fiber optic cable, or twisted pair, then the coaxial cable, fiber optic cable, or twisted pair are included in the definition of medium. Disk and disc, as used herein, include compact disc (CD), laser disc, optical disc, digital versatile disc (DVD), and Blu-Ray® disc where disks usually reproduce data magnetically, while discs reproduce data optically with lasers. Combinations of the above should also be included within the scope of computer-readable media.

[0172] In other words, like a human driver of a vehicle, the autonomous vehicle 100 can be programmed to follow traffic laws and rules of the road, and to make decisions based on its programming and the information it receives from its sensors and other sources. The autonomous vehicle 100 can also be programmed to make decisions based on its programming and the information it receives from its sensors and other sources, even if those decisions are not in accordance with traffic laws and rules of the road. For example, the autonomous vehicle 100 can be programmed to avoid a collision with another vehicle, even if doing so would violate a traffic law or rule of the road.

[0173] In other words, like a human driver of a vehicle, the autonomous vehicle 100 can be programmed to follow traffic laws and rules of the road, and to make decisions based on its programming and the information it receives from its sensors and other sources. The autonomous vehicle 100 can also be programmed to make decisions based on its programming and the information it receives from its sensors and other sources, even if those decisions are not in accordance with traffic laws and rules of the road. For example, the autonomous vehicle 100 can be programmed to avoid a collision with another vehicle, even if doing so would violate a traffic law or rule of the road.

[0174] It should be understood that aspects of the application can be implemented in hardware, software, firmware or combinations thereof. In the embodiments described above, various steps or methods can be implemented, for example, by software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, and in another embodiment, any of the following technology, known in the art, or combinations thereof, can be used: discrete logic circuitry having logic gates for implementing logic functions upon an application of data signals, application specific integrated circuits having appropriate combinational logic gates, programmable gate arrays (PGA), field programmable gate arrays (FPGA), and the like.

[0175] It should be noted that the above examples are only used to illustrate the technical solutions of the present application but not limit the present application. Although the present application has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present application can be modified or replaced by equivalents without departing from the spirit and scope of the present application, and all of them should be covered in the scope of the claims of the present application.

Claims

1. A greenhouse environment remote monitoring method applied to greenhouse planting, characterized in that, The method comprises the following steps: S1, building a sensor network to collect environmental parameters; Establish a sensor network based on the Internet of Things to collect environmental parameters in the greenhouse, and obtain environmental data; S2, preprocessing the environmental data; Correct the environmental data through data preprocessing technology to obtain cleaned data; S3, multi-dimensional environmental analysis based on cleaned data; Use principal component analysis technology to identify the relationship between environmental data in the greenhouse, and obtain the analysis result; S4, environmental modeling based on the analysis result; Based on the multi-layer perceptron MLP network structure model, a prediction model of the environment in the greenhouse and crop growth is constructed, and an environmental model is obtained; The multi-layer perceptron MLP network structure model comprises: Establish an environmental model of the environmental parameters in the greenhouse on crop growth: ； Wherein, the crop growth state output at all times ; for inputting values of environmental parameters at time t, including temperature , humidity , air flow speed , light intensity and carbon dioxide concentration ; are preset values representing time-varying weight coefficients of each input parameter, the relative influence of each environmental factor on crop growth at different time points; bias, which represents the base output value of the network, is initialized as 0; For the nonlinear activation function, its function expression is ; Introduce the time sequence weighting mechanism: Introduce the time window parameter, so that each input not only depends on the data at the current time, but also depends on the environmental data in the history time, which is represented by the following formula: ； Wherein, is the length of the time window, the crop growth state at the current time instant depends not only on the environmental data at the current time instant, but also on the environmental data at past time instants time instants; representing in the past a weighted sum of the momentary intra- environmental parameters; for the past crop growth state output within the time instant; S5, temperature and humidity prediction and adjustment suggestion based on the environmental model; Based on the long short-term memory network LSTM, the temperature and humidity of the environmental model are predicted and adjustment suggestions are generated, and the adjustment suggestions are obtained; S6, automatic control based on the adjustment suggestion; Use the automatic control system based on the industrial automation communication protocol Modbus protocol to realize remote control of the environmental adjustment equipment in the greenhouse, and obtain the automatic control result; S7, real-time environmental parameter collection based on the automatic control result, and real-time feedback of the collection result.

2. The greenhouse environment remote monitoring method applied to greenhouse planting according to claim 1, characterized in that, In step S1, the environmental parameters include: Temperature and humidity, air flow speed, light intensity and carbon dioxide concentration.

3. The method of claim 1, wherein the method is applied to remote monitoring of the environment in a greenhouse. In step S2, the data preprocessing technology includes: Data interpolation: use the mean, median or mode of the environmental data to fill in the missing data; Outlier detection: calculate the standard score of the environmental data point, if the difference between the environmental data point and the mean is greater than twice or three times the standard deviation, it is considered that the point is abnormal; Data smoothing: reduce noise by calculating the average value in the window around the environmental data point.

4. The greenhouse environment remote monitoring method for greenhouse planting according to claim 1, characterized in that, In step S3, the principal component analysis technology includes: Find the direction with the largest variance in the cleaned data to extract the main features, and the environmental parameter with the largest variance in the cleaned data has the greatest impact on crop growth.

5. The method of claim 1, wherein the method is applied to remote monitoring of the environment in a greenhouse for plant cultivation. In step S4, the multi-layer perceptron MLP network structure model further comprises: Temperature and humidity prediction and analysis.

6. The greenhouse environment remote monitoring method applied to greenhouse planting according to claim 5, characterized in that, The temperature and humidity prediction and analysis includes: Data collection: obtain the cleaned data and analysis results, and record them in time sequence; Model training: input the collected data, use the multi-layer perceptron MLP network structure model to train, and learn the nonlinear relationship between the input environmental factors and the temperature and humidity change; Temperature and humidity prediction: after training, use the new real-time environmental data including the current temperature and humidity, air flow speed and light intensity to obtain the temperature and humidity prediction result at the future time; Environmental adjustment suggestion: according to the predicted temperature and humidity prediction result, adjust the temperature and humidity control equipment in the greenhouse to maintain a suitable crop growth environment.

7. The method of claim 1, wherein the method is applied to remote monitoring of an indoor environment of a greenhouse. In step S5, the long short-term memory network LSTM includes: Use the multi-layer perceptron MLP network structure model to obtain the environmental prediction model in the greenhouse; Adopting multi-dimension joint processing, all environment factors are jointly modeled and an environment prediction model is used, and an activation function is used for each environment factor , and the activation function is a memory unit of a long short-term memory network (LSTM). The time window weighting mechanism is adopted, that is, the historical data in each time window is weighted according to the influence degree of the environmental factor, and the influence of the environmental factor on the crop growth environment is determined by calculating the variance, and the greater the variance, the greater the influence.

8. The greenhouse environment remote monitoring method for greenhouse planting according to claim 1, characterized in that, In the step S7, the sensor network established in the step S1 collects the environmental parameters in the greenhouse in real time, compares the difference between the adjustment result of the equipment after the automatic control and the actual environmental parameters, and provides a reference basis for the adjustment of the equipment.

Images