Underforest planting intelligent environment monitoring system and method

Abstract

The invention provides an under-forest planting intelligent environment monitoring system and method, and the system comprises a sensor arrangement subsystem which is used for arranging a sensor cluster; the under-forest monitoring data acquisition subsystem is used for acquiring under-forest monitoring data through a sensor cluster; the intelligent sprinkling irrigation control subsystem is used for performing intelligent sprinkling irrigation according to the under-forest monitoring data based on a deep learning technology; and the intelligent early warning subsystem is used for carrying out environment intelligent early warning according to the under-forest monitoring data based on a fuzzy control technology. According to the intelligent environment monitoring system and method for under-forest planting, the intelligent sprinkling irrigation model based on deep learning is constructed, the switching time of the sprinkling irrigation system can be determined according to the meteorological conditions and the current soil and air humidity, the soil and air humidity suitable for growth of under-forest plants is achieved, and the sprinkling irrigation control process is more intelligent; and an intelligent environment early warning model is constructed, so that a user can be reminded of the growth condition of the plants, and the production efficiency and the product quality of the plants planted under the forest are improved.

Description

Technical Field

[0001] The present invention relates to the field of smart agricultural technology, and in particular to an intelligent environment monitoring system and method for understory planting. Background Art

[0002] Smart agriculture is a new agricultural model that applies modern information technology to agricultural production. It integrates sensors, the Internet of Things, big data analysis, and artificial intelligence to accurately monitor and manage the agricultural production process, thereby improving agricultural production efficiency, reducing costs, improving product quality, and achieving sustainable development.

[0003] At present, the level of intelligence (intelligent control) of under-forest planting needs to be improved. For example, how to effectively and accurately identify the production environment conditions of planting products (such as Ganoderma lucidum), how to judge the time and duration of sprinkler irrigation, etc. These problems still need to be solved, which to a certain extent limits production efficiency and product quality.

[0004] In view of this, there is an urgent need for an intelligent environmental monitoring system and method for understory planting to at least solve the above-mentioned deficiencies. Summary of the invention

[0005] One of the purposes of the present invention is to provide an intelligent environmental monitoring system and method for understory planting, and to construct an intelligent sprinkler irrigation model based on deep learning. The system can determine the switching time of the sprinkler irrigation system according to meteorological conditions and current soil and air humidity, so as to achieve soil and air humidity suitable for the growth of understory plants, and make the sprinkler irrigation control process more intelligent; an intelligent environmental early warning model can be constructed to help remind users of the growth status of plants and improve the production efficiency and product quality of understory plants.

[0006] The intelligent environment monitoring system for understory planting provided by the embodiment of the present invention includes:

[0007] A sensor deployment subsystem, used to deploy sensor clusters;

[0008] The understory monitoring data acquisition subsystem is used to conduct intelligent understory planting environment monitoring and acquire understory monitoring data through sensor clusters;

[0009] Intelligent sprinkler control subsystem, which is used to perform intelligent sprinkler irrigation based on deep learning technology and forest monitoring data;

[0010] The intelligent early warning subsystem is used to provide intelligent environmental early warning based on the forest understory monitoring data based on fuzzy control technology.

[0011] Preferably, the sensor deployment subsystem includes:

[0012] A location determination module is provided to determine the on-site upwind location of the target planting base for understory planting;

[0013] The deployment module is used to deploy a sensor cluster at an upwind location on site; wherein the deployed sensor cluster includes: a light sensor, an air temperature and humidity sensor, a carbon dioxide sensor, a soil temperature sensor, and a soil moisture sensor.

[0014] Preferably, the forest understory monitoring data acquisition subsystem includes:

[0015] The understory monitoring data acquisition module is used to obtain understory monitoring data through a sensor cluster based on a preset monitoring cycle and a cloud server.

[0016] Preferably, the intelligent sprinkler irrigation control subsystem comprises:

[0017] Intelligent irrigation model building module, used to build intelligent irrigation models based on deep learning technology;

[0018] Sprinkler control module, used for intelligent sprinkler irrigation based on intelligent irrigation model and forest monitoring data;

[0019] Among them, the intelligent irrigation model building module includes:

[0020] The first training submodule is used to collect historical monitoring environment indicators during non-irrigation periods, capture the temporal changes of historical monitoring environments during non-irrigation periods, and train an irrigation start model;

[0021] The second training submodule is used to collect historical monitoring environment indicators during irrigation, capture the temporal changes of historical monitoring environment during irrigation, and train the irrigation shutdown model;

[0022] An integration module, for obtaining an intelligent irrigation model according to an irrigation on model and an irrigation off model;

[0023] Among them, the sprinkler control module includes:

[0024] The irrigation start time determination submodule is used to use the LSTM model to perform multi-time step prediction output when irrigation is not turned on. When the predicted relative humidity of the air reaches the preset first rule condition, the length of the first time step is counted to determine the irrigation start time;

[0025] The irrigation shut-off time determination submodule is used to use the LSTM model to perform multi-time step prediction output when irrigation is turned on. When the predicted relative humidity of the air reaches the preset second rule condition, the length of the second time step is counted to determine the irrigation shut-off time.

[0026] Preferably, the intelligent irrigation model building module further performs the following operations:

[0027] Obtain the first artificial cultivation record of the target planting base;

[0028] Based on a preset cultivation feature extraction template, extract a first artificial cultivation feature; the first artificial cultivation feature includes: a first irrigation timing, a first irrigation parameter and a first irrigation effect;

[0029] Screening the first artificial cultivation record whose first irrigation effect is greater than or equal to a preset irrigation effect threshold and using it as the first target artificial cultivation record;

[0030] Connect to the big data supplement node to obtain the second artificial cultivation record of the same species as the understory plants;

[0031] Based on the cultivation feature extraction template, extracting a second artificial cultivation feature; the second artificial cultivation feature includes: a second irrigation timing, a second irrigation parameter, and a second irrigation effect;

[0032] Screening the second artificial cultivation records whose second irrigation effect is greater than or equal to a preset irrigation effect threshold and using them as the second target artificial cultivation records;

[0033] Performing a first suitability analysis on the irrigation environment at the second irrigation time, and determining whether the first suitability analysis is passed;

[0034] Performing a second suitability analysis on the second irrigation parameter, and determining whether the second suitability analysis passes;

[0035] If both the first suitability analysis and the second suitability analysis are passed, the corresponding second target artificial cultivation record will be used as the third target artificial cultivation record;

[0036] Based on deep learning technology, the intelligent irrigation model is trained according to the first target artificial cultivation records and the third target artificial cultivation records;

[0037] The second suitability analysis is performed on the second irrigation parameter, and judging whether the second suitability analysis is passed includes:

[0038] determining, according to the irrigation capacity of the irrigation device, whether the second irrigation parameter can be achieved;

[0039] If this can be achieved, the second suitability analysis is passed.

[0040] Preferably, performing a first suitability analysis on the irrigation environment at the second irrigation opportunity and determining whether the first suitability analysis passes includes:

[0041] Acquire first environmental characteristic data at a first irrigation time, and determine environmental characteristic data intervals of different first environmental characteristic types according to the first environmental characteristic data;

[0042] Acquire second environmental characteristic data at a second irrigation opportunity;

[0043] Determining a matching interval of the second environmental feature data in the environmental feature data interval according to the second environmental feature type of the second environmental feature data;

[0044] Determine whether the second environmental feature data falls within the matching interval, and if so, obtain a first target weight of the first environmental feature type corresponding to the matching interval;

[0045] If not, obtaining a second target weight of the first environmental feature type corresponding to the matching interval;

[0046] Get the data distribution of the matching interval;

[0047] Based on the percentile method and data distribution, determine the benchmark interval value of the matching interval;

[0048] Determine the difference between the second environment characteristic data and the reference interval value, and calculate the absolute value of the difference between the difference value and the length of the matching interval;

[0049] According to the quotient of the absolute value and the length of the matching interval, the weight of the second target is lowered;

[0050] Determining the environmental feature reproducibility according to the sum of the first target weight corresponding to the second environmental feature type and / or the second target weight after being adjusted down;

[0051] If the environmental feature reproducibility is greater than or equal to a preset environmental feature reproducibility threshold, the first suitability analysis is passed.

[0052] Preferably, the intelligent early warning subsystem includes:

[0053] Membership function design module, used to design membership functions based on the growth environment requirements of understory planting targets;

[0054] A fuzzy quantification result determination module is used to perform fuzzy quantization on the forest understory monitoring data according to the membership function and determine the fuzzy quantization result;

[0055] A knowledge rule base construction module is used to construct a knowledge rule base based on expert experience and planting experience;

[0056] A fuzzy reasoning module is used to perform fuzzy reasoning based on the fuzzy quantification results and the knowledge rule base to obtain fuzzy output quantities;

[0057] The fuzzy warning module is used to make fuzzy judgments and fuzzy warnings based on the fuzzy output.

[0058] The embodiment of the present invention provides an intelligent environment monitoring method for understory planting, comprising:

[0059] Step 1: Deploy sensor clusters;

[0060] Step 2: Use sensor clusters to conduct intelligent environmental monitoring of understory planting and obtain understory monitoring data;

[0061] Step 3: Based on deep learning technology and forest monitoring data, intelligent sprinkler irrigation is carried out;

[0062] Step 4: Based on fuzzy control technology, intelligent environmental warning is carried out according to the forest understory monitoring data.

[0063] Preferably, step 1: deploying a sensor cluster includes:

[0064] Determine the on-site upwind location of the target planting site for understory planting;

[0065] A sensor cluster is deployed at an upwind location on site; wherein the deployed sensor cluster includes: a light sensor, an air temperature and humidity sensor, a carbon dioxide sensor, a soil temperature sensor, and a soil moisture sensor.

[0066] Preferably, step 2: using a sensor cluster to perform intelligent environmental monitoring of understory planting and obtain understory monitoring data, including:

[0067] Based on the preset monitoring cycle and cloud server, forest monitoring data is obtained through sensor clusters.

[0068] Preferably, step 3: based on deep learning technology, intelligent sprinkler irrigation is performed according to forest monitoring data, including:

[0069] Build intelligent irrigation models based on deep learning technology;

[0070] Based on the intelligent irrigation model, intelligent sprinkler irrigation is carried out according to the forest monitoring data;

[0071] Among them, based on deep learning technology, an intelligent irrigation model is built, including:

[0072] Collect historical monitoring environment indicators during non-irrigation periods, capture the temporal changes of historical monitoring environments during non-irrigation periods, and train irrigation start-up models;

[0073] Collect historical monitoring environment indicators during irrigation, capture the temporal changes of historical monitoring environment during irrigation, and train irrigation shutdown models;

[0074] According to the irrigation on model and the irrigation off model, an intelligent irrigation model is obtained;

[0075] Among them, based on the intelligent irrigation model and according to the forest monitoring data, intelligent sprinkler irrigation is carried out, including:

[0076] When irrigation is not turned on, the LSTM model is used to perform multi-time step prediction output. When the predicted relative humidity of the air reaches the preset first rule condition, the length of the first time step is counted to determine the irrigation start time;

[0077] When irrigation is turned on, the LSTM model is used to perform multi-time-step prediction output. When the predicted relative humidity of the air reaches the preset second rule condition, the length of the second time step is counted to determine the time to turn off irrigation.

[0078] Preferably, step 4: Based on fuzzy control technology, according to forest monitoring data, intelligent environmental warning is performed, including:

[0079] Design membership functions based on the growth environment requirements of understory planting targets;

[0080] According to the membership function, the understory monitoring data is fuzzy quantified to determine the fuzzy quantification result;

[0081] Build a knowledge rule base based on expert experience and planting experience;

[0082] Perform fuzzy reasoning based on fuzzy quantification results and knowledge rule base to obtain fuzzy output;

[0083] Make fuzzy judgment and fuzzy warning based on fuzzy output.

[0084] The beneficial effects of the present invention are:

[0085] The present invention constructs an intelligent sprinkler irrigation model based on deep learning, which can determine the switching time of the sprinkler irrigation system according to meteorological conditions, current soil and air humidity, so as to achieve soil and air humidity suitable for the growth of understory plants, and make the sprinkler irrigation control process more intelligent; constructing an intelligent environmental early warning model can help remind users of the growth status of plants and improve the production efficiency and product quality of understory plants.

[0086] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0087] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0088] Figure 1 Schematic diagram of an intelligent environment monitoring system for understory planting in an embodiment of the present invention;

[0089] Figure 2 It is a fuzzy control principle diagram of the intelligent environment monitoring system for understory planting in an embodiment of the present invention;

[0090] Figure 3It is a viewing applet interface of the understory planting intelligent environment monitoring system in the embodiment of the present invention;

[0091] Figure 4 It is a design diagram of the remote control system for automatic monitoring and sprinkler irrigation of the understory planting intelligent environment in the embodiment of the present invention;

[0092] Figure 5 Schematic diagram of an intelligent environmental monitoring method for understory planting in an embodiment of the present invention. DETAILED DESCRIPTION

[0093] The preferred embodiments of the present invention are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0094] The embodiment of the present invention provides an intelligent environment monitoring system for understory planting, such as Figure 1 As shown, including:

[0095] The sensor deployment subsystem 1 is used to deploy sensor clusters; wherein the sensor clusters are: sensors deployed in the understory planting area, used to collect environmental data of the deployment area;

[0096] The understory monitoring data acquisition subsystem 2 is used to perform understory planting intelligent environmental monitoring through a sensor cluster and acquire understory monitoring data; the understory monitoring data includes: light data, air temperature and humidity data, carbon dioxide concentration data, soil temperature data, and soil humidity data;

[0097] Intelligent sprinkler irrigation control subsystem 3, used for intelligent sprinkler irrigation based on deep learning technology and forest monitoring data; wherein, intelligent sprinkler irrigation based on deep learning technology and forest monitoring data is: deep learning a large amount of sprinkler irrigation data of forest plants (such as Ganoderma lucidum), automatically adjusting the operation of the sprinkler irrigation system according to the collected forest monitoring data, and realizing precise irrigation;

[0098] The intelligent early warning subsystem 4 is used to carry out intelligent environmental early warning based on the forest monitoring data based on fuzzy control technology. Among them, based on the fuzzy control principle, according to the growth environment requirements of forest plants for light intensity, temperature and humidity at different growth periods, based on appropriate environmental thresholds, combined with the characteristics of the growth period, different membership functions are designed, including Gaussian membership functions, trapezoidal membership functions and triangular membership functions, and knowledge bases and rule bases are established, so as to adopt different early warning times for environmental indicators during the growth period. The fuzzy control principle is as follows: Figure 2 shown.

[0099] The working principle and beneficial effects of the above technical solution are:

[0100] The construction of the automatic monitoring system for understory planting environment and remote intelligent monitoring system for sprinkler irrigation includes two parts of work:

[0101] 1) Deployment of automatic monitoring system for understory planting environment and remote control system for sprinkler irrigation;

[0102] 2) Development of an intelligent monitoring system for understory planting environment based on deep learning and fuzzy control.

[0103] 1) Layout of automatic monitoring of forest planting environment and remote control system for sprinkler irrigation

[0104] 1.1) Layout of automated atomization sprinkler irrigation system

[0105] With high-pressure atomizing sprinkler irrigation equipment, filters and control boxes as the hardware foundation, a full-coverage atomizing sprinkler irrigation system is built. For example, according to the spacing width of Ganoderma lucidum planting, the spacing width of water mist sprinkler nozzles (1.2m) is arranged, and according to the humidity requirement range of Ganoderma lucidum, the height of water mist sprinkler nozzles is arranged. After the water source is filtered and pressurized, high-pressure atomizing sprinkler irrigation is realized to ensure the high humidity requirement of the planting environment; combined with intelligent valves, GPRS mobile communication technology, cloud servers, and remote control technology, data information is transmitted back to the cloud data center through the intelligent gateway, and managers can remotely view and control the on-site sprinkler irrigation facilities through WEB terminals or mobile phone APPs. The system supports preset time and operation cycle. When the preset conditions are reached, the system will automatically operate the on-site equipment to complete the control action.

[0106] 1.2) Deployment of automatic monitoring stations based on smart environment

[0107] The atmospheric environment information and soil environment information of forest planting are monitored in real time through sensor clusters and automatic weather stations. An automatic environmental monitoring station is set upwind of the planting base. The monitoring station is composed of light, air temperature and humidity, carbon dioxide, soil temperature, and soil humidity sensors. The monitoring station transmits environmental information data to the forest intelligent monitoring system server through the 5G network and cloud server every 1 minute.

[0108] 2) Development of an intelligent monitoring system for understory planting environment based on deep learning and fuzzy control

[0109] 2.1) Intelligent sprinkler irrigation control model based on LSTM

[0110] By applying artificial intelligence and deep learning algorithms, we establish the response relationship between soil moisture, air humidity and sprinkler irrigation duration / pressure under different meteorological conditions (light, temperature, carbon dioxide, etc.), and construct an intelligent sprinkler irrigation prediction model. The intelligent sprinkler irrigation prediction model can determine the switching time of the sprinkler irrigation system based on meteorological conditions and current soil and air humidity to achieve soil and air humidity suitable for understory planting.

[0111] The intelligent irrigation model is divided into two parts, including irrigation on and irrigation off models. The irrigation on model collects environmental indicators monitored by the understory planting base during the non-irrigation period, and captures the temporal changes in the growth environment of understory plants during the non-irrigation period. The irrigation off model collects environmental indicators monitored by the understory planting base during the irrigation period, and captures the temporal changes in the growth environment of understory plants during the irrigation period.

[0112] Taking the prediction of the main control indicator relative humidity as an example, considering that the current relative humidity is affected by related indicators within a time series, the intelligent irrigation model uses the currently widely used LSTM time series model to predict the relative humidity of the air of understory plants.

[0113] The basic computational unit of the LSTM model is as follows:

[0114] I t =σ(X t W xi +H t-1 W hi +b i )

[0115] F t =σ(X t W xf +H t-1 W hf +b f )

[0116] O t =σ(X t W xo +H t-1 W ho +b o )

[0117]

[0118] H t =O t ⊙tanh(C t )

[0119] Y=H t ⊙W xy +b y

[0120] Among them, I t 、F t , O t They are the input gate, forget gate, and output gate features at time t, respectively. C t They are candidate memory cells and memory cells at time t, respectively. xi , W hi , W xf , Whf , W xo , W ho , W xc , W hc , W xy , b i , b f , b o , b c , b y are model parameters, X t is the input data feature, which is the monitoring environment indicator feature at time t in the time series. t-1 , H t is the hidden state. Y is the prediction index, which is the relative humidity of the air at time t+1, which is predicted by feature learning in the time series.

[0121] Based on the LSTM regression model, the relative humidity of the air can be predicted in multiple time steps. When irrigation is not turned on, the LSTM regression model is used to perform multi-time step prediction output. When the predicted relative humidity of the air reaches the preset rule conditions (unsuitable growth environment), the time step length is counted to obtain the irrigation time. Similarly, during the period of irrigation, the LSTM regression model is used to perform multi-time step prediction output. When the predicted relative humidity of the air reaches the preset rule conditions (suitable growth conditions), the time step length is counted to obtain the irrigation shutdown time.

[0122] 2.2) Environmental intelligent early warning model based on fuzzy control algorithm

[0123] Combining expert experience and field planting experience, we can clarify the range of light intensity, temperature and humidity suitable for understory plants at different growth periods. We can apply fuzzy decision-making algorithms to build an intelligent environmental early warning model. We can conduct intelligent early warning and forecasting for scenarios where environmental factors exceed the appropriate range, and remind managers to take corresponding measures.

[0124] Based on the principle of fuzzy control, different membership functions are designed according to the growth environment requirements of understory plants for light intensity, temperature and humidity in different growth periods. These include Gaussian membership function, trapezoidal membership function and triangular membership function. A knowledge base and a rule base are established to adopt different warning times for environmental indicators in the growth period, thereby reducing the false alarm rate and interference frequency.

[0125] 2.3) Integrated understory planting environment intelligent control system

[0126] The forest planting intelligent monitoring sprinkler irrigation system is implemented in the form of SQL database + model + applet, which is convenient for users to log in and access anytime and anywhere. The applet can be used to remotely view the on-site planting environment. Figure 3At the same time, the system is based on the planting environment intelligent control model and early warning model algorithm to achieve real-time automatic control of the sprinkler system and risk warning of growth environment parameters exceeding the standard. The design diagram of the intelligent automatic monitoring and sprinkler remote control system for understory planting is shown in the figure. Figure 4 shown.

[0127] The system backend database is built with mainstream SQL data to store the monitoring indicators of the understory planting environment in real time. To avoid the loss of monitoring data, the system determines abnormal data in real time and sends the abnormal data to the administrator. The judgment logic is: ① Continuous missing for 1 hour; or ② More than 20% of the data is missing in a day, that is, the number of missing minutes accounts for more than 20% of the total number of minutes in a day; or ③ The soil humidity, air humidity, and carbon dioxide have a value of 0 for 1 hour; or ④ More than 20% of the data in a day of soil humidity, air humidity, and carbon dioxide are 0 values.

[0128] The system consists of five modules. From top to bottom, they are Module 1: Display the switch status of sprinkler equipment and key monitoring indicators (soil moisture); Module 2: Display the real-time value of monitoring indicators; Module 3: Display the time series value of monitoring indicators, and select time and indicators through the filter button; Module 4: Display the current working condition of sprinkler equipment; Module 5: Click the "Start Equipment" button to perform remote irrigation operations. Click the "Abnormal Data" button to view abnormal data judged by the system.

[0129] The present invention constructs an intelligent sprinkler irrigation model based on deep learning, which can determine the switching time of the sprinkler irrigation system according to meteorological conditions, current soil and air humidity, so as to achieve soil and air humidity suitable for the growth of understory plants, and make the sprinkler irrigation control process more intelligent; constructing an intelligent environmental early warning model can help remind users of the growth status of plants and improve the production efficiency and product quality of understory plants.

[0130] In one embodiment, the sensor deployment subsystem includes:

[0131] The layout position determination module is used to determine the on-site upwind position of the target planting base for understory planting; wherein the target planting base is: the planting base corresponding to the understory plants;

[0132] The deployment module is used to deploy a sensor cluster at an upwind location on site; wherein the deployed sensor cluster includes: a light sensor, an air temperature and humidity sensor, a carbon dioxide sensor, a soil temperature sensor, and a soil moisture sensor.

[0133] The working principle and beneficial effects of the above technical solution are:

[0134] The present invention uses a sensor cluster and an automatic weather station to monitor the atmospheric environment information, soil environment information and other environmental information of understory planting in real time, and deploys an automatic environmental monitoring station at the upwind location of the planting base, thereby improving the rationality of the sensor location deployment.

[0135] In one embodiment, the understory monitoring data acquisition subsystem includes:

[0136] The understory monitoring data acquisition module is used to acquire understory monitoring data through a sensor cluster based on a preset monitoring cycle and a cloud server. The preset monitoring cycle is manually set, for example, 1 minute.

[0137] The working principle and beneficial effects of the above technical solution are:

[0138] The monitoring station in the present invention transmits environmental information data to the forest intelligent monitoring system server through the 5G network and cloud server every 1 minute, so that data monitoring is more timely.

[0139] In one embodiment, the intelligent sprinkler irrigation control subsystem includes:

[0140] Intelligent irrigation model building module, used to build intelligent irrigation models based on deep learning technology;

[0141] Sprinkler control module, used for intelligent sprinkler irrigation based on intelligent irrigation model and forest monitoring data;

[0142] Among them, the intelligent irrigation model building module includes:

[0143] The first training submodule is used to collect historical monitoring environment indicators during non-irrigation periods, capture the temporal changes of historical monitoring environments during non-irrigation periods, and train an irrigation start model;

[0144] The second training submodule is used to collect historical monitoring environment indicators during irrigation, capture the temporal changes of historical monitoring environment during irrigation, and train the irrigation shutdown model;

[0145] An integration module, for obtaining an intelligent irrigation model according to an irrigation on model and an irrigation off model;

[0146] Among them, the sprinkler control module includes:

[0147] The irrigation start time determination submodule is used to use the LSTM model to perform multi-time step prediction output when irrigation is not turned on. When the predicted relative humidity of the air reaches the preset first rule condition, the length of the first time step is counted to determine the irrigation start time; wherein the first rule condition is: the environment is not suitable for the growth of understory plants; the length of the first time step is: the time step from the current time to the predicted irrigation start time;

[0148] The irrigation closing time determination submodule is used to use the LSTM model to perform multi-time step prediction output when irrigation is turned on. When the predicted relative humidity of the air reaches the preset second rule condition, the second time step length is counted to determine the irrigation closing time. Among them, the second rule condition is: suitable growth conditions for understory plants; the second time step length is: the time step from the current time to the predicted irrigation closing time.

[0149] The working principle and beneficial effects of the above technical solution are:

[0150] The present invention uses artificial intelligence and deep learning algorithms to establish the response relationship between soil moisture, air humidity and sprinkler irrigation time / pressure under different meteorological conditions (light, temperature, carbon dioxide, etc.), and construct an intelligent sprinkler irrigation prediction model. The intelligent sprinkler irrigation prediction model can determine the switch time of the sprinkler irrigation system according to meteorological conditions and current soil and air humidity to achieve soil and air humidity suitable for understory planting;

[0151] The intelligent irrigation model is divided into two parts, including irrigation on and irrigation off models. The irrigation on model collects environmental indicators monitored by the target planting base during the non-irrigation period and captures the temporal changes in the growth environment of understory plants during the non-irrigation period. The irrigation off model collects environmental indicators monitored by the target planting base during the irrigation period and captures the temporal changes in the growth environment of understory plants during the irrigation period.

[0152] Taking the prediction of the main control indicator of relative air humidity as an example, considering that the current relative air humidity is affected by related indicators within a time series, the intelligent irrigation model uses the currently widely used LSTM time series model to predict the relative air humidity of understory plants;

[0153] Based on the LSTM regression model, the relative humidity of the air in multiple time steps can be predicted. When irrigation is not turned on, the LSTM regression model is used to perform multi-time step prediction output. When the predicted relative humidity of the air reaches the preset rule conditions (unsuitable growth environment), the time step length at this time is counted to obtain the irrigation time. Similarly, during the period when irrigation is turned on, the LSTM regression model is used to perform multi-time step prediction output. When the predicted relative humidity of the air reaches the preset rule conditions (suitable growth conditions), the time step length is counted to obtain the irrigation shutdown time.

[0154] In one embodiment, the intelligent irrigation model building module further performs the following operations:

[0155] Obtain the first artificial cultivation record of the target planting base; wherein the first artificial cultivation record is: the record of artificial cultivation of understory plants in the target planting base in history, such as: watering record;

[0156] Based on the preset cultivation feature extraction template, the first artificial cultivation feature is extracted; the first artificial cultivation feature includes: a first irrigation time, a first irrigation parameter and a first irrigation effect; wherein the preset cultivation feature extraction template is: a template for extracting artificial cultivation features by comparing with artificial cultivation records; the irrigation time is: environmental parameters when irrigation occurs; the irrigation parameter is: watering amount; the irrigation effect is: a comparative score between the growth state of understory plants and the average growth state between watering and the next watering time;

[0157] Screening the first artificial cultivation record whose first irrigation effect is greater than or equal to a preset irrigation effect threshold value and using it as the first target artificial cultivation record; wherein the preset irrigation effect threshold value is manually preset;

[0158] Connect to the big data supplement node to obtain the second artificial cultivation record of the same species as the understory plants; the big data supplement node is: understory plant planting exchange platform;

[0159] Based on the cultivation feature extraction template, extracting a second artificial cultivation feature; the second artificial cultivation feature includes: a second irrigation timing, a second irrigation parameter, and a second irrigation effect;

[0160] Screening the second artificial cultivation records whose second irrigation effect is greater than or equal to a preset irrigation effect threshold and using them as the second target artificial cultivation records;

[0161] Performing a first suitability analysis on the irrigation environment at the second irrigation time to determine whether the first suitability analysis is passed; wherein the first suitability analysis is to determine whether the second irrigation time is suitable for local application;

[0162] Performing a second suitability analysis on the second irrigation parameter to determine whether the second suitability analysis is passed; wherein the second suitability analysis is to determine whether the second irrigation parameter is suitable for local application;

[0163] If both the first suitability analysis and the second suitability analysis are passed, the corresponding second target artificial cultivation record will be used as the third target artificial cultivation record;

[0164] Based on deep learning technology, the intelligent irrigation model is trained according to the first target artificial cultivation records and the third target artificial cultivation records;

[0165] The second suitability analysis is performed on the second irrigation parameter, and judging whether the second suitability analysis is passed includes:

[0166] According to the irrigation capacity of the irrigation device, judging whether the second irrigation parameter can be achieved; wherein the irrigation device is an irrigation device used in the target planting base; when judging whether the second irrigation parameter can be achieved, judging whether the second irrigation parameter can be achieved according to the irrigation capacity;

[0167] If this can be achieved, the second suitability analysis is passed.

[0168] The working principle and beneficial effects of the above technical solution are:

[0169] The present invention introduces two types of artificial cultivation records for subsequent training of intelligent irrigation models; the first type is the first artificial cultivation record of the target planting base, which is more applicable because the first artificial cultivation record is directly obtained from the base; a cultivation feature extraction template is introduced to determine the first irrigation time, the first irrigation parameter and the first irrigation effect, and the first target artificial cultivation record whose first irrigation effect is greater than or equal to the preset irrigation effect threshold is screened; the second type is to connect the big data supplement node to obtain the second artificial cultivation record of the same type as the understory plants, extract the second irrigation time, the second irrigation parameter and the second irrigation effect based on the cultivation feature extraction template, and determine the second target artificial cultivation record whose second irrigation effect is greater than or equal to the preset irrigation effect threshold; considering that the irrigation mechanism and growth environment of the understory plants corresponding to the second artificial cultivation record obtained by big data may be greatly different from those of the target planting base, the first suitability analysis is performed on the second irrigation time and the second suitability analysis is performed on the second irrigation parameter, and the third target artificial cultivation record suitable for local irrigation model training is screened; based on the deep learning technology, the intelligent irrigation model is trained according to the first target artificial cultivation record and the third target artificial cultivation record, thereby improving the comprehensiveness and rationality of the irrigation model training.

[0170] In one embodiment, performing a first suitability analysis on the irrigation environment at the second irrigation opportunity and determining whether the first suitability analysis passes includes:

[0171] Acquire first environmental characteristic data of a first irrigation opportunity, and determine environmental characteristic data intervals of different first environmental characteristic types according to the first environmental characteristic data; wherein the first environmental characteristic data is environmental data corresponding to the first irrigation opportunity, such as light intensity, air temperature and humidity, carbon dioxide content, and soil temperature and humidity; the environmental characteristic data intervals of different first environmental characteristic types are: the value range of the first environmental characteristic data of the first environmental characteristic type determined by summarizing the first environmental characteristic data of the same first environmental characteristic type, such as: taking the minimum value obtained by summarization as the minimum value of the interval, and taking the maximum value obtained by summarization as the maximum value of the interval;

[0172] Acquire second environmental characteristic data at a second irrigation opportunity; wherein the second environmental characteristic data is environmental data corresponding to the second irrigation opportunity;

[0173] Determine, according to the second environment feature type of the second environment feature data, a matching interval of the second environment feature data in the environment feature data interval; wherein the first environment feature type corresponding to the matching interval is consistent with the second environment feature type of the second environment feature data corresponding to the matching interval;

[0174] Determine whether the second environmental feature data falls within the matching interval, and if so, obtain a first target weight of the first environmental feature type corresponding to the matching interval; wherein the first target weight of the first environmental feature type is manually preset;

[0175] If not, obtaining a second target weight of the first environmental feature type corresponding to the matching interval;

[0176] Obtaining data distribution of the matching interval; wherein the data distribution is: which first environmental feature data is in which interval position of the matching interval;

[0177] Based on the percentile method and according to the data distribution, the benchmark interval value of the matching interval is determined; wherein, based on the percentile method and according to the data distribution, the benchmark interval value of the matching interval is determined by: determining the upper and lower bounds of the matching interval according to specific percentiles (for example, 2.5% and 97.5%), obtaining a benchmark interval containing 95% of the reference individuals, and calculating the average value of the first environmental characteristic data within the benchmark interval to obtain the benchmark interval value;

[0178] Determine the difference between the second environment characteristic data and the reference interval value, and calculate the absolute value of the difference between the difference value and the length of the matching interval;

[0179] According to the quotient of the absolute value and the length of the matching interval, the second target weight is lowered; wherein, when the second target weight is lowered, the quotient and the second target weight are correspondingly multiplied;

[0180] Calculate the sum of the first target weight and / or the lowered second target weight corresponding to the second environmental feature type to determine the environmental feature reproducibility;

[0181] If the environmental feature reproducibility is greater than or equal to a preset environmental feature reproducibility threshold, the first suitability analysis is passed, wherein the preset environmental feature reproducibility threshold is manually preset.

[0182] The working principle and beneficial effects of the above technical solution are:

[0183] The present invention obtains first environmental characteristic data at a first irrigation time, performs data statistics, determines an environmental characteristic data interval corresponding to each first environmental characteristic type; determines a second environmental characteristic type of the second environmental characteristic data, determines a matching interval corresponding to the environmental characteristic data interval of the second environmental characteristic type; determines whether the second environmental characteristic data falls within the matching interval, if the second environmental characteristic data falls within the matching interval, it indicates that the environmental data reuse degree of the first environmental characteristic type corresponding to the matching interval is high, and determines a first target weight of the corresponding first environmental characteristic type; if the second environmental characteristic data does not fall within the matching interval, obtains the matching interval. For the second target weight corresponding to the first environmental feature type, the percentile method is introduced to determine the benchmark interval value according to the data distribution of the matching interval; according to the difference between the second environmental feature data and the benchmark interval value, the downward adjustment weight of the second target weight (the quotient of the absolute value and the length of the matching interval) is calculated, and the second target weight is downward adjusted; the sum of the first target weight corresponding to the second environmental feature type and / or the downward adjusted second target weight is used as the environmental feature reproducibility, and when the environmental feature reproducibility is greater than or equal to the preset environmental feature reproducibility threshold, it is determined that the first suitability analysis has passed, thereby improving the accuracy of the first suitability analysis process.

[0184] In one embodiment, the intelligent early warning subsystem includes:

[0185] Membership function design module, used to design membership functions based on the growth environment requirements of understory planting targets;

[0186] A fuzzy quantification result determination module is used to perform fuzzy quantization on the forest understory monitoring data according to the membership function and determine the fuzzy quantization result;

[0187] A knowledge rule base construction module is used to construct a knowledge rule base based on expert experience and planting experience;

[0188] A fuzzy reasoning module is used to perform fuzzy reasoning based on the fuzzy quantification results and the knowledge rule base to obtain fuzzy output quantities;

[0189] The fuzzy warning module is used to make fuzzy judgments and fuzzy warnings based on the fuzzy output.

[0190] The working principle and beneficial effects of the above technical solution are:

[0191] The present invention combines expert experience and field planting experience to clarify the range of light intensity, temperature and humidity suitable for understory plants at different growth periods; applies fuzzy decision-making algorithms to build an intelligent environmental early warning model; and conducts intelligent early warning and forecasting for scenarios where environmental factors exceed the appropriate range, reminding administrators to take corresponding measures;

[0192] Based on the principle of fuzzy control, different membership functions are designed according to the growth environment requirements of understory plants in different growth periods for light intensity, temperature and humidity, based on appropriate environmental thresholds and combined with the characteristics of the growth period, including Gaussian membership function, trapezoidal membership function and triangular membership function. A knowledge base and a rule base are established, so that different warning times are adopted for the environmental indicators in the growth period, thereby reducing the false alarm rate and interference frequency.

[0193] The embodiment of the present invention provides an intelligent environmental monitoring method for understory planting, such as Figure 5 As shown, including:

[0194] Step 1: Deploy sensor clusters;

[0195] Step 2: Use sensor clusters to conduct intelligent environmental monitoring of understory planting and obtain understory monitoring data;

[0196] Step 3: Based on deep learning technology and forest monitoring data, intelligent sprinkler irrigation is carried out;

[0197] Step 4: Based on fuzzy control technology, intelligent environmental warning is carried out according to the forest understory monitoring data.

[0198] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. The intelligent environmental monitoring system for understory planting is characterized by: include: A sensor deployment subsystem, used to deploy sensor clusters; The understory monitoring data acquisition subsystem is used to conduct intelligent understory planting environment monitoring and acquire understory monitoring data through sensor clusters; Intelligent sprinkler control subsystem, which is used to perform intelligent sprinkler irrigation based on deep learning technology and forest monitoring data; The intelligent early warning subsystem is used to provide intelligent environmental early warning based on the forest understory monitoring data based on fuzzy control technology.

2. The intelligent environment monitoring system for understory planting according to claim 1, characterized in that: Sensor deployment subsystem, including: A location determination module is provided to determine the on-site upwind location of the target planting base for understory planting; The deployment module is used to deploy a sensor cluster at an upwind location on site; wherein the deployed sensor cluster includes: a light sensor, an air temperature and humidity sensor, a carbon dioxide sensor, a soil temperature sensor, and a soil moisture sensor.

3. The intelligent environment monitoring system for understory planting according to claim 1, characterized in that: The understory monitoring data acquisition subsystem includes: The understory monitoring data acquisition module is used to obtain understory monitoring data through a sensor cluster based on a preset monitoring cycle and a cloud server.

4. The intelligent environment monitoring system for understory planting according to claim 1, characterized in that: Intelligent sprinkler control subsystem, including: Intelligent irrigation model building module, used to build intelligent irrigation models based on deep learning technology; Sprinkler control module, used for intelligent sprinkler irrigation based on intelligent irrigation model and forest monitoring data; Among them, the intelligent irrigation model building module includes: The first training submodule is used to collect historical monitoring environment indicators during non-irrigation periods, capture the temporal changes of historical monitoring environments during non-irrigation periods, and train an irrigation start model; The second training submodule is used to collect historical monitoring environment indicators during irrigation, capture the temporal changes of historical monitoring environment during irrigation, and train the irrigation shutdown model; An integration module, for obtaining an intelligent irrigation model according to an irrigation on model and an irrigation off model; Among them, the sprinkler control module includes: The irrigation start time determination submodule is used to use the LSTM model to perform multi-time step prediction output when irrigation is not turned on. When the predicted relative humidity of the air reaches the preset first rule condition, the length of the first time step is counted to determine the irrigation start time; The irrigation shut-off time determination submodule is used to use the LSTM model to perform multi-time step prediction output when irrigation is turned on. When the predicted relative humidity of the air reaches the preset second rule condition, the length of the second time step is counted to determine the irrigation shut-off time.

5. The intelligent environment monitoring system for understory planting according to claim 1, characterized in that: Intelligent early warning subsystem, including: Membership function design module, used to design membership functions based on the growth environment requirements of understory planting targets; A fuzzy quantification result determination module is used to perform fuzzy quantization on the forest understory monitoring data according to the membership function and determine the fuzzy quantization result; A knowledge rule base construction module is used to construct a knowledge rule base based on expert experience and planting experience; A fuzzy reasoning module is used to perform fuzzy reasoning based on the fuzzy quantification results and the knowledge rule base to obtain fuzzy output quantities; The fuzzy warning module is used to make fuzzy judgments and fuzzy warnings based on the fuzzy output.

6. An intelligent environmental monitoring method for understory planting, characterized in that: include: Step 1: Deploy sensor clusters; Step 2: Use sensor clusters to conduct intelligent environmental monitoring of understory planting and obtain understory monitoring data; Step 3: Based on deep learning technology and forest monitoring data, intelligent sprinkler irrigation is carried out; Step 4: Based on fuzzy control technology, intelligent environmental warning is carried out according to the forest understory monitoring data.

7. The method for monitoring the understory planting environment according to claim 6, characterized in that: Step 1: Deploy the sensor cluster, including: Determine the on-site upwind location of the target planting site for understory planting; A sensor cluster is deployed at an upwind location on site; wherein the deployed sensor cluster includes: a light sensor, an air temperature and humidity sensor, a carbon dioxide sensor, a soil temperature sensor, and a soil moisture sensor.

8. The method for monitoring the understory planting environment according to claim 6, characterized in that: Step 2: Use the sensor cluster to conduct intelligent environmental monitoring of forest planting and obtain forest monitoring data, including: Based on the preset monitoring cycle and cloud server, forest monitoring data is obtained through sensor clusters.

9. The method for monitoring the intelligent environment of understory planting according to claim 6, characterized in that: Step 3: Based on deep learning technology and forest monitoring data, intelligent sprinkler irrigation is performed, including: Build intelligent irrigation models based on deep learning technology; Based on the intelligent irrigation model, intelligent sprinkler irrigation is carried out according to the forest monitoring data; Among them, based on deep learning technology, an intelligent irrigation model is built, including: Collect historical monitoring environment indicators during non-irrigation periods, capture the temporal changes of historical monitoring environments during non-irrigation periods, and train irrigation start-up models; Collect historical monitoring environment indicators during irrigation, capture the temporal changes of historical monitoring environment during irrigation, and train irrigation shutdown models; According to the irrigation on model and the irrigation off model, an intelligent irrigation model is obtained; Among them, based on the intelligent irrigation model and according to the forest monitoring data, intelligent sprinkler irrigation is carried out, including: When irrigation is not turned on, the LSTM model is used to perform multi-time step prediction output. When the predicted relative humidity of the air reaches the preset first rule condition, the length of the first time step is counted to determine the irrigation start time; When irrigation is turned on, the LSTM model is used to perform multi-time-step prediction output. When the predicted relative humidity of the air reaches the preset second rule condition, the length of the second time step is counted to determine the time to turn off irrigation.

10. The method for monitoring the understory planting environment according to claim 6, characterized in that: Step 4: Based on fuzzy control technology and forest monitoring data, intelligent environmental warning is carried out, including: Design membership functions based on the growth environment requirements of understory planting targets; According to the membership function, the understory monitoring data is fuzzy quantified to determine the fuzzy quantification result; Build a knowledge rule base based on expert experience and planting experience; Perform fuzzy reasoning based on fuzzy quantification results and knowledge rule base to obtain fuzzy output; Make fuzzy judgment and fuzzy warning based on fuzzy output.

Images