An automated irrigation method and apparatus for high yield cultivation of potatoes

Abstract

The present application relates to the technical field of vegetable cultivation, and particularly relates to an automatic irrigation method and device for high-yield cultivation of potatoes, comprising: constructing a growth function of potatoes in different growth periods to determine the image acquisition frequency of potatoes in each preset period, thereby obtaining a plurality of potato images; constructing a neural network for identifying the maturity value of potatoes to obtain the maturity value of potatoes corresponding to each underground water outlet pipe at the current time; constructing a maturity value input into a maturity value-soil humidity requirement distribution function to determine the soil humidity requirement value corresponding to each underground water outlet pipe at the current time; and combining the comparison result of the size of the soil humidity value at each water outlet of each underground water outlet pipe at the current time to obtain the electromagnetic valve switching result of each water outlet of each underground water outlet pipe at the current time. The present application can realize precise drip irrigation by regulating the electromagnetic valves of the water outlets at different depths of the underground water outlet pipes.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of vegetable cultivation, in particular to an automatic irrigation method and device for high-yield cultivation of potatoes. BACKGROUND

[0002] To increase the yield of potatoes and achieve high yield, different cultivation methods are often tried, such as ridge cultivation, mulching cultivation, and nest cultivation, but precise and reasonable field management can also achieve the purpose of high yield of potatoes.

[0003] Existing problems: In the past, irrigation methods often depended on human experience to irrigate according to the growth state of potatoes and the soil moisture condition (soil moisture condition refers to the soil moisture condition), and the irrigation was performed as appropriate. This method relies on human experience, and improper watering can easily cause water resource waste, cause weak growth of potato seedlings or overgrowth, and result in low yield and poor potato quality. The rise of automation and intelligent technology in the information age has brought about the demand and exploration of agricultural standardization technology and process, and the realization of precise irrigation of potatoes is a technical problem that needs to be solved. SUMMARY

[0004] The present application provides an automatic irrigation method and device for high-yield cultivation of potatoes to solve the existing problems.

[0005] The automatic irrigation method and device for high-yield cultivation of potatoes of the present application adopt the following technical solutions:

[0006] An embodiment of the present application provides an automatic irrigation method for high-yield cultivation of potatoes, which comprises the following steps:

[0007] Obtain the soil moisture value at each time at each water outlet of each underground water outlet pipe in the potato planting field;

[0008] Construct a growth function of potatoes in different growth periods, determine the potato image acquisition frequency in each preset period according to the size of the output value of the growth function of potatoes in different growth periods input in each preset period, and obtain a plurality of potato images according to the potato image acquisition frequency in each preset period; wherein each potato image corresponds to an underground water outlet pipe and a time;

[0009] Construct a neural network for identifying potato maturity values, obtain a trained neural network, input all the potato images corresponding to each underground water outlet pipe at the current time into the trained neural network, and obtain the potato maturity value corresponding to each underground water outlet pipe at the current time;

[0010] According to the size of the potato maturity value corresponding to each underground water outlet pipe at the current moment, the soil humidity demand value corresponding to each underground water outlet pipe at the current moment is determined; according to the comparison result of the soil humidity demand value corresponding to each underground water outlet pipe at the current moment and the soil humidity value at each water outlet of each underground water outlet pipe at the current moment, the electromagnetic valve switching result of each water outlet of each underground water outlet pipe at the current moment is obtained.

[0011] Further, the growth function of the constructed potato in different growth periods includes the following specific steps:

[0012] The duration of the potato in different growth periods is obtained, and the growth function of the potato in the germination period, the seedling period, the tuber formation period, the tuber formation peak period and the tuber formation end period is constructed according to the duration of the potato in different growth periods using the membership function.

[0013] Further, the specific steps of determining the potato image acquisition frequency in each preset period include the following:

[0014] Any one preset period is input into the growth function of the potato in different growth periods, and a plurality of probability values are output, and the acquisition frequency in the arbitrary one preset period is determined according to the size of the maximum probability value.

[0015] Further, the specific steps of determining the acquisition frequency in the arbitrary one preset period according to the size of the maximum probability value include the following:

[0016] The upward rounding value of the product of the maximum probability value and the preset basic acquisition frequency is taken as the acquisition frequency in the arbitrary one preset period.

[0017] Further, the specific steps of constructing the neural network for identifying the maturity value of the potato include the following:

[0018] The neural network for identifying the maturity value of the potato includes a data set, a final label of each potato image, and a loss function.

[0019] The data set is composed of all potato images.

[0020] The loss function is constructed according to the final label.

[0021] Further, the final label of each potato image is composed of the growth period classification value corresponding to each potato image and the maturity value.

[0022] Further, the specific steps of obtaining the potato maturity value corresponding to each underground water outlet pipe at the current moment include the following:

[0023] Input each potato image into the trained neural network to obtain an output maturity value of each potato image;

[0024] Take the average of the output maturity values of all potato images corresponding to each underground outlet pipe at the current time as the potato maturity value corresponding to each underground outlet pipe at the current time.

[0025] Further, the specific steps of determining the soil humidity requirement value corresponding to each underground outlet pipe at the current time include the following:

[0026] A potato maturity value-soil humidity requirement distribution function is constructed, and the potato maturity value corresponding to each underground outlet pipe at the current time is input into the potato maturity value-soil humidity requirement distribution function to output the soil humidity requirement value corresponding to each underground outlet pipe at the current time.

[0027] Further, the specific steps of obtaining the electromagnetic valve switching result of each outlet of each underground outlet pipe at the current time include the following:

[0028] If the soil humidity requirement value corresponding to each underground outlet pipe at the current time is greater than the soil humidity value corresponding to each outlet of each underground outlet pipe at the current time, the electromagnetic valve of each outlet of each underground outlet pipe is opened.

[0029] If the soil humidity requirement value corresponding to each underground outlet pipe at the current time is less than or equal to the soil humidity value corresponding to each outlet of each underground outlet pipe at the current time, the electromagnetic valve of each outlet of each underground outlet pipe is closed.

[0030] The application further provides an automatic irrigation equipment for high-yield potato cultivation, which comprises a memory, a processor, and a computer program stored in the memory and capable of running on the processor.

[0031] The technical scheme of the application has the following beneficial effects:

[0032] In the embodiment of the present application, a corresponding irrigation system is arranged in the potato planting field, a growth function of the potato in different growth periods is constructed, the image acquisition frequency of the potato in each preset period is determined, the image of the potato field is acquired through the camera erected beside, and thus the image acquisition amount of the potato in different growth periods is ensured to be sufficient through the adaptive image acquisition frequency, and the effect of subsequent neural network training is improved. A neural network for identifying the maturity value of the potato is constructed, the maturity degree is identified and quantified through the neural network, that is, the maturity value of the potato corresponding to each underground water outlet pipe at the current time is acquired, and thus the accuracy of the acquired maturity value of the potato is ensured. The maturity value of the potato is input into a maturity value-soil humidity requirement distribution function in combination with the soil humidity requirement amount of different growth periods, the soil humidity requirement value of each underground water outlet pipe at the current time is determined, the electromagnetic valve switching result of each water outlet of each underground water outlet pipe at the current time is acquired in combination with the comparison result of the size of the soil humidity value at each water outlet of each underground water outlet pipe at the current time. Thus, the electromagnetic valve of the water outlet of different depths of the underground water outlet pipe can be regulated according to the real-time soil humidity information, and precise drip irrigation is realized. BRIEF DESCRIPTION OF DRAWINGS

[0033] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or the prior art description will be briefly introduced. 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.

[0034] Figure 1 A step flow chart of the automatic irrigation method for high-yield cultivation of potatoes of the present application;

[0035] Figure 2 A schematic diagram of the automatic irrigation system for high-yield cultivation of potatoes;

[0036] Figure 3 A schematic diagram of the overall structure of the underground water outlet pipe;

[0037] Figure 4 A schematic diagram of the overall internal structure of the underground water outlet pipe;

[0038] Figure 5 A schematic diagram of the parallel expansion group structure of the underground water outlet pipe;

[0039] Figure 6 A schematic diagram of the parallel expansion group structure of the underground water outlet pipe from the top view;

[0040] Figure 7 A schematic diagram of a humidity sensor;

[0041] Figure 8 a growth function diagram of potatoes in different growth periods;

[0042] Figure 9 a superimposed growth function diagram of potatoes in all growth periods;

[0043] Figure 10 a marking diagram of potatoes in different growth periods;

[0044] Figure 11 a neural network structure diagram;

[0045] Figure 12 a potato maturity value-soil humidity requirement distribution curve diagram. DETAILED DESCRIPTION

[0046] In order to further clarify the technical means and effects taken by the present application to achieve the predetermined object of the application, the specific implementation, structure, features and effects of the automatic irrigation method and equipment for high-yield cultivation of potatoes according to the present application are described in detail as follows. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. In addition, the specific features, structures or characteristics in one or more embodiments can be combined in any suitable form.

[0047] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the present application belongs.

[0048] The specific scheme of the automatic irrigation method and equipment for high-yield cultivation of potatoes according to the present application is described in detail below in conjunction with the accompanying drawings.

[0049] Please refer to Figure 1 which shows a step flowchart of the automatic irrigation method for high-yield cultivation of potatoes provided by one embodiment of the present application, which includes the following steps:

[0050] Step S001: Obtain the soil humidity value at each time at each water outlet of each underground water outlet pipe in the potato planting field.

[0051] In this embodiment, the matching device is used for underground drip irrigation, combined with a humidity sensor, and data is transmitted through the Internet of Things. The server is responsible for receiving data and performing analysis and calculation (analyzing the current potato growth period and calculating the corresponding soil humidity requirement). According to the analysis data, the electromagnetic valve device is adjusted, and the irrigation control is performed through the on-off state of the electromagnetic valve. The device has a monitoring function and can irrigate at different depths, making it more accurate. At the same time, it meets the irrigation needs of deep plowing, a high-yield method, and automatically identifies the growth state, allowing timely adjustment of soil moisture. Specifically: In the potato field, the corresponding irrigation system is arranged, the camera beside it collects the image of the potato field, and the neural network is used to identify and quantify the maturity, combined with the soil humidity requirement of different growth periods, the soil humidity requirement of the potato tuber area is automatically obtained, and according to the real-time soil humidity information, the electromagnetic valve of the water outlet at different depths is adjusted to achieve precise drip irrigation.

[0052] It should be noted that in this embodiment, the existing water outlet pipe is improved, and an electronic sensor is configured, which can supply water to soil at different depths. The cam structure allows the sensor to be inserted horizontally into the soil to measure soil humidity. The irrigation method uses the membrane drip irrigation technology, specifically, the water outlet pipe is vertically inserted into the ground, and different drip irrigation outlets are distributed in the longitudinal depth direction of the water outlet pipe to achieve drip irrigation of soil at different depths. At the same time, humidity sensors, distance sensors, Bluetooth communication modules, solar charging modules, etc. are installed. The device has both monitoring and control functions. The entire irrigation equipment includes: a water storage tank for storing water for drip irrigation. A water delivery pipe for delivering water from the water storage tank to each water outlet pipe. An above-ground water outlet pipe for implementing ground irrigation, which normally follows seasonal irrigation and is not regulated in this embodiment. An underground water outlet pipe for monitoring soil humidity at different depths and implementing drip irrigation at different depths, which requires real-time monitoring and regulation. A visual detection device for monitoring and collecting the status image of the entire potato planting area. An Internet of Things terminal that obtains sensor Bluetooth communication data and transmits it to the cloud server through WiFi wireless transmission. A cloud server for data analysis and calculation. The software part of the server includes: a growth cycle fuzzy judgment module that fuzzy judges the growth state of potatoes based on the planting time. A maturity visual reasoning module that collects potato ground images and reasons the maturity value through a neural network. A soil humidity requirement calculation module that calculates the corresponding soil humidity requirement value based on the maturity value. An automatic irrigation system for high-yield potato cultivation is shown in Figure 2 , Figure 2 which includes: a growth cycle fuzzy judgment module, a maturity visual reasoning module, a soil humidity requirement calculation module, a real-time soil humidity monitoring module, and an electromagnetic valve control module.

[0053] It needs to be further explained that the water storage barrel is arranged beside the field, connected through the pipeline at the bottom and the water delivery pipe, and the water delivery pipe is connected with the above-ground water outlet pipe and the underground water outlet pipe. The underground water outlet pipe is provided with a solar panel at the top, which is used to collect solar energy and convert it into electric energy to power the electronic module above. The bottom is a tapered structure, which is convenient for insertion into the soil. A distance measuring sensor is installed on the side, which is used to determine the depth of the water outlet pipe inserted into the soil, and then identify the depth of each drip irrigation water outlet. The top of the underground water outlet pipe has an interface for connecting with the water delivery pipe. The water in the water delivery pipe is divided into multiple branch pipes for drip irrigation at different depths. At the same time, a multi-layer cam structure is installed inside for opening multiple water outlets and humidity sensors. The data collected by the sensor is transmitted through the built-in Bluetooth module, and control information is also received through the Bluetooth sensor, so as to control the electromagnetic valves of each water outlet pipe. The overall structure diagram of the underground water outlet pipe is shown in Figure 3 . The overall internal structure diagram of the underground water outlet pipe is shown in Figure 4 . The parallel expansion group structure diagram of the underground water outlet pipe is shown in Figure 5 . The top view diagram of the parallel expansion group structure of the underground water outlet pipe is shown in Figure 6 . Figure 3 , Figure 4 , Figure 5 and Figure 6 1 is a solar panel, 2 is an interface, 3 is a water outlet pipe barrel, 4 is an expansion roller, 5 is a parallel expansion group, 6 is a sharp cone, 7 is a cam, 8 is a connecting wheel, 9 is a support rod, 10 is a humidity sensor, and 11 is a drip irrigation water outlet. The support rod has a spring. Deep plowing (deeper for potato planting) is a way to effectively increase potato yield, so the depth of the underground water outlet pipe should be at least 70 cm or more to adapt to different cultivation methods. The drip irrigation water outlets can be evenly arranged on the underground water outlet pipe. The humidity sensor diagram is shown in Figure 7 . The humidity sensor is inserted into the soil to detect the soil humidity. The collection frequency of soil humidity is once per second. In this embodiment, a humidity sensor is arranged at each water outlet of the same underground water outlet pipe in the longitudinal direction (perpendicular to the ground). Each underground water outlet pipe has multiple water outlets in the longitudinal direction.

[0054] In the potato planting field, the soil humidity value at each water outlet of each underground water outlet pipe at each moment is obtained.

[0055] It needs to be explained that the real-time collected soil humidity data of each humidity sensor is subjected to mean filtering and denoising processing, which is a known technology, and the specific method is not introduced here, which improves the accuracy of data analysis.

[0056] Step S002: constructing a growth function of the potato in different growth periods, determining a potato image acquisition frequency in each preset period according to a size of an output value input into the growth function of the potato in different growth periods in each preset period, and acquiring a plurality of potato images according to the potato image acquisition frequency in each preset period, wherein each potato image corresponds to an underground water pipe and a time point.

[0057] It should be noted that the above is the corresponding device for irrigation, and a corresponding method is also needed to control the above device to realize accurate irrigation of the potato. The growth of the potato mainly includes four stages: plant growth, tuber formation, tuber growth and maturity period, wherein the main purpose is to ensure sufficient water during the tuber growth period and to allow the tuber to grow as much as possible, thereby achieving high yield. At the same time, the water needs to be controlled moderately during the maturity period to improve the dry matter content. The potato is a crop that requires a lot of water, and from the tuber formation period to the peak tuber formation period, reasonable irrigation needs to be performed according to the soil moisture condition (the soil moisture condition refers to the soil humidity condition) to keep the relative humidity of the soil between 65% and 75%, preferably between 67% and 73%, and more preferably between 69% and 71%.

[0058] It should be further noted that the judgment of the maturity degree of the potato in the embodiment is mainly through two stages, that is, a fuzzy estimation is first realized through the planting time, and then the neural network technology is used to identify the maturity value according to the real-time field image. The specific processing process of obtaining the soil humidity requirement of the potato is as follows: (1) the growth period of the potato is roughly judged through the growth time. (2) The camera collects the field image, and the neural network technology is used to identify the maturity value of the potato. (3) The soil humidity requirement corresponding to the maturity value of the potato is set. (4) The humidity requirement of the tuber region of the potato under the field coordinates is obtained. Since the sowing time of the potato in Inner Mongolia is usually in the middle and late May, and the harvest period is from August to October, the total growth period of the potato is about 80 to 120 days (early-maturing varieties are 80 days, and late-maturing varieties are 120 days). Among them, the germination period lasts for about 10 to 20 days, the seedling period lasts for about 15 to 20 days, the tuber formation period lasts for about 20 to 30 days, the peak tuber formation period lasts for about 15 to 25 days, and the late tuber formation period lasts for about 15 to 25 days.

[0059] Preferably, in an embodiment of the present application, the method for acquiring the potato image acquisition frequency in each preset period comprises:

[0060] In the embodiment, the membership function is used to construct the growth function of the potato in different growth periods according to the duration of the potato in different growth periods.

[0061] It should be noted that the three common membership functions are triangle, trapezoidal and Gaussian, which are known techniques and will not be described in detail. In this embodiment, the growth function of potatoes in the germination period , the growth function of potatoes in the seedling period , the growth function of potatoes in the tuber formation period , the growth function of potatoes in the peak tuber formation period , the growth function of potatoes in the late tuber formation period , wherein, is the independent variable (days) in the growth function of potatoes in different growth periods, and the growth function of potatoes in different growth periods is shown in Figure 8 , and Figure 8 , from top to bottom, are the growth function diagrams of potatoes in the germination period, the seedling period, the tuber formation period, the peak tuber formation period and the late tuber formation period. In this embodiment, the preset period is 1 day, which is taken as an example for description. Figure 8 , the horizontal axis is time, the unit is day, the planting day is 0, the time range is 0 to 130 days, and the vertical axis is probability value, the probability range is between 0 and 1. If the probability value corresponding to any day in the growth function diagram of potatoes in the seedling period is larger, it means that the day is more likely to be in the seedling period of potatoes. The superimposed diagram of the growth function of potatoes in all growth periods is shown in Figure 9 . Figure 9 The horizontal and vertical axes in Figure 8 are the same as

[0062] The preset basic collection frequency is 2000 per day, which is taken as an example for description.

[0063] Input any preset period into the growth function of potatoes in different growth periods, output several probability values, and take the upward integer value of the product of the maximum probability value and the preset basic collection frequency as the collection frequency in the arbitrary preset period.

[0064] It should be noted that, for example, the 62nd day has probability values of the tuber formation period, the peak tuber formation period and the late tuber formation period, and the maximum probability is the peak tuber formation period. Therefore, it is considered that the 62nd day is the peak tuber formation period of potatoes, and then the frequency of potato image collection is regulated according to the probability value of the 62nd day in the peak tuber formation period of potatoes, which is used for high-frequency image collection to judge more accurate potato maturity, and also serves as an abnormality judgment standard for growth degree. If the neural network inference result is no longer in the range of probability obtained in the above growth function curve, it is determined that the neural network recognition is abnormal.

[0065] According to the acquisition frequency in each preset period, a plurality of potato images are acquired; each potato image corresponds to an underground water outlet pipe and a time point.

[0066] It should be noted that according to the acquisition frequency of each day, all cameras arranged in the potato field are used to collect the images of the potato field every day, and each potato image corresponds to a time stamp. The position and shooting angle of each camera are fixed, so the shooting range of each camera is fixed. Since the drip irrigation range of each underground water outlet pipe in the potato field is fixed, the shooting range of each camera can be divided into the drip irrigation ranges of different underground water outlet pipes by manual marking, so that the images taken by each camera are divided into image blocks corresponding to the drip irrigation ranges of different underground water outlet pipes, which are used as potato images. Therefore, each potato image corresponds to an underground water outlet pipe.

[0067] Step S003: constructing a neural network for identifying the maturity value of potato, acquiring the trained neural network; inputting all the potato images corresponding to each underground water outlet pipe at the current time into the trained neural network to acquire the maturity value of potato corresponding to each underground water outlet pipe at the current time.

[0068] It should be noted that in different growth periods of potato, the branches, leaves and flowers of the plant have different performances, and the maturity value of potato can be determined by the specific performance of the plant. However, when multiple factors are used to determine the maturity value, specific determination conditions need to be set for each factor, and this method also depends on professional planting and machine vision knowledge. In this embodiment, the complete growth images of the potato in the field are collected, and the growth cycle of the key time point is labeled to generate continuous maturity value labels. After training the neural network, the input image can output the corresponding maturity value.

[0069] Preferably, in one embodiment of the present application, the method for acquiring the maturity value of potato corresponding to each underground water outlet pipe at the current time comprises:

[0070] In this embodiment, the construction of the neural network for identifying the maturity value of potato includes data set acquisition, label making, network construction, network training and network use.

[0071] Therefore, the cameras are arranged around the planted potato field to shoot images of the field from multiple angles in time sequence. The multiple-angle images of multiple fields ensure that the raw image data is greater than 100,000, i.e. the data set is constructed by using the above-mentioned potato images.

[0072] Then, based on experience, continuous images are manually identified and judged, with images clearly marked as boundaries between growth stages. For example, seedling emergence marks the boundary between germination and seedling stages; budding marks the boundary between seedling and tuber formation; full bloom marks the boundary between tuber formation and tuber growth; and senescent and yellowing of stems and leaves marks the boundary between tuber growth and the final stage of tuber formation. To quantify potato maturity, the numerical space from 0 to 1 is divided into five intervals, corresponding to the germination stage, seedling stage, tuber formation stage, peak tuber formation stage, and final tuber formation stage. The marked points are the boundaries for each stage. The process of obtaining the quantitative values ​​for the dividing points is as follows: Based on the maximum growth time, a timeline of potato growth status is obtained. Specifically, for the 20-day germination period, a label value of 0.2 is assigned to the first 19 days after planting; for the 20-day seedling period, a label value of 0.4 is assigned to days 20 to 39; for the 30-day tuber formation period, a label value of 0.6 is assigned to days 40 to 69; for the 25-day peak tuber formation period, a label value of 0.8 is assigned to days 70 to 94; and for the 25-day late tuber formation period, a label value of 1 is assigned to days 95 to 120. This is used to construct the timeline. The dividing time between the germination and seedling stages is day 20; the dividing time between the seedling and tuber formation stages is day 40; the dividing time between the tuber formation and peak tuber formation stages is day 70; and the dividing time between the peak tuber formation and late tuber formation stages is day 75. A schematic diagram of potato labels at different growth stages is shown below. Figure 10 As shown. Figure 10 The horizontal and vertical axes represent time, with the unit being days. The value is 0 for the day of planting, and the time range is from 0 to 125 days.

[0073] From the day the potatoes were planted until day 120, a straight line was fitted using the least squares method based on the label values ​​for all days. The horizontal axis of the fitted line represents time, and the vertical axis represents the label values. The label value at the corresponding time point for each potato image was obtained from the fitted line and used as the maturity value for each image. The least squares method is a well-known technique, and its specific method will not be described here.

[0074] Simultaneously, based on the range of maturity values, the classification value of the growth stage corresponding to each potato image is obtained. Figure 10 On the timeline, if each potato image corresponds to any growth stage, then the label value for that growth stage is set to 1, and the label values ​​for all other growth stages are set to 0. The label values ​​for the classification task are only 0 and 1. Therefore, the classification values ​​for the growth stage corresponding to each potato image can be divided into 5 categories, which are: Category 1: Category 2: Category 3: Category 4: Category 5: ,in , , , and are the label values of the sprouting stage, seedling stage, tuber formation stage, peak tuber formation stage and end tuber formation stage corresponding to each potato image, respectively.

[0075] Thus, according to the growth stage classification value and the maturity value corresponding to each potato image, the final label of each potato image is formed .

[0076] It should be noted that in order to ensure more accurate results, a two-branch neural network structure is adopted, which needs to integrate classification and regression tasks, and at the same time realizes the output of specific maturity values while classifying the potato images into 5 categories. The classification task corresponds to 5 growth state intervals, and the numerical regression function outputs a value between 0 and 1, which is used to represent the growth state of the potato plant in the image. The network first uses the alternating method of convolutional layer and pooling layer to realize the collection of image features and the dimension reduction of data features, realizing the encoding function of image data, and then realizes classification and data regression through a fully connected network. As for the loss function of the network, the cross-entropy function needs to be selected for the classification part, and the mean square error loss function is selected for the data regression part, which is a known technology and the specific method is not introduced here. The neural network structure diagram is shown in Figure 11 . Figure 11 The image enters the encoder and gets the classification part and the regression part through the fully connected network. The loss function of the current structure neural network is:

[0077] For the classification part, the output values of the first to fifth neurons of the output layer (i.e. the label values of the sprouting stage, seedling stage, tuber formation stage, peak tuber formation stage and end tuber formation stage of the potato ) are processed by the sigmoid activation function in the neuron, and the result value is a decimal between 0 and 1, so the loss function corresponding to the classification part is:

[0078]

[0079] In the formula, is the label value of the th growth stage corresponding to each potato image, takes the value of 0 or 1, represent the sprouting stage, seedling stage, tuber formation stage, peak tuber formation stage and end tuber formation stage of the potato, respectively. the output value of the th neuron. is the logarithmic function.

[0080] For the regression part, the loss function of the output value of the sixth neuron (i.e. corresponding to the mature value ) of the output layer is:

[0081]

[0082] In the formula, is the output value of the 6th neuron.

[0083] The preset first weight is 0.4, and the preset second weight is 0.6. This is described as an example.

[0084] In this embodiment, in addition to the basic classification and regression requirements, the loss function

[0085] Thus, the construction of the neural network for identifying the mature value of the potato is completed.

[0086] The network training is: through setting the training parameters such as the maximum training times, the training batch, the learning rate and the like, the network is trained through multiple rounds of training and gradually converges. The neural network training is completed when the loss function converges, approaches to zero and is stable. Thus, the trained neural network is obtained.

[0087] The use of the network is: after the neural network training is completed, the parameters are fixed and no longer adjusted, and the image to be inferred can be directly input into the neural network to obtain the corresponding inference value. The form of the inference value is: , which respectively represent the output values of the 1st, 2nd, 3rd, 4th, 5th and 6th neurons. The construction process of the growth function of the potato in different growth periods through the fuzzy results obtained by time can be monitored for abnormalities. That is, if the inference result of the neural network is that the current growth period is the second growth cycle (seedling stage), and the probability value of the day number of the second growth cycle is determined through the planting time, the result of the neural network is recognized, otherwise an abnormal prompt is given.

[0088] Thus, all the potato images corresponding to each underground water pipe at the current time are output to the trained neural network to obtain the mature value of the potato corresponding to each underground water pipe at the current time.

[0089] It should be noted that: in this embodiment, each potato image is input into the trained neural network for identifying the mature value of the potato to obtain the output mature value of each potato image. The mean value of the output mature values of all the potato images corresponding to each underground water pipe at the current time is taken as the mature value of the potato corresponding to each underground water pipe at the current time.

[0090] ​​Step S004: determining the soil humidity demand value corresponding to each underground water outlet pipe at the current time according to the size of the potato maturity value corresponding to each underground water outlet pipe at the current time; and obtaining the electromagnetic valve switching result of each water outlet of each underground water outlet pipe at the current time according to the comparison result of the soil humidity demand value corresponding to each underground water outlet pipe at the current time and the soil humidity value at each water outlet of each underground water outlet pipe at the current time.

[0091] It should be noted that the soil humidity requirement of each period is artificially set, and the soil humidity requirement and the label value (maturity) in the fitting straight line are one-to-one corresponding at the same time. The least square method is used for curve fitting to obtain a curve with maturity as the horizontal axis and soil humidity as the vertical axis, and a potato maturity value-soil humidity requirement distribution function corresponding to the curve. The potato maturity value-soil humidity requirement distribution curve is shown in Figure 12 Figure 12 The horizontal axis is the maturity value, and the vertical axis is the soil humidity requirement.

[0092] Preferably, in an embodiment of the present application, the method for obtaining the electromagnetic valve switching result of each water outlet of each underground water outlet pipe at the current time comprises:

[0093] The potato maturity value corresponding to each underground water outlet pipe at the current time is input into the potato maturity value-soil humidity requirement distribution function, and the soil humidity demand value corresponding to each underground water outlet pipe at the current time is output.

[0094] If the soil humidity demand value corresponding to each underground water outlet pipe at the current time is greater than the soil humidity value corresponding to each water outlet of each underground water outlet pipe at the current time, the electromagnetic valve of each water outlet of each underground water outlet pipe is opened for drip irrigation. If the soil humidity demand value corresponding to each underground water outlet pipe at the current time is less than or equal to the soil humidity value corresponding to each water outlet of each underground water outlet pipe at the current time, the electromagnetic valve of each water outlet of each underground water outlet pipe is closed, and drip irrigation is not performed. In this way, the drip irrigation frequency is adjusted to achieve more accurate irrigation control.

[0095] It should be noted that due to the small fluctuations in humidity sensor data, the average value of the soil humidity value at each water outlet of each underground water outlet pipe in the last half hour at the current time is taken as the soil humidity value at each water outlet of each underground water outlet pipe at the current time. This is an example.

[0096] ​The application further provides an automatic irrigation equipment for high-yield cultivation of potatoes, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor executes the computer program stored in the memory to realize the steps of the automatic irrigation method for high-yield cultivation of potatoes.

[0097] Thus far, the application is completed.

[0098] To sum up, in the embodiment of the application, the growth function of potatoes in different growth periods is constructed to determine the image acquisition frequency of potatoes in each preset period, so as to obtain a plurality of potato images, the neural network for identifying the maturity value of potatoes is constructed to obtain the maturity value of potatoes corresponding to each underground water outlet pipe at the current time, the maturity value of potatoes is input into the maturity value-soil humidity requirement distribution function to determine the soil humidity requirement value corresponding to each underground water outlet pipe at the current time, and the comparison result of the size of the soil humidity value at each water outlet of each underground water outlet pipe at the current time is obtained to obtain the electromagnetic valve switching result of each water outlet of each underground water outlet pipe at the current time. The application can realize precise drip irrigation by regulating the electromagnetic valves of the water outlets of the underground water outlet pipes at different depths.

[0099] The above only describes the preferred embodiments of the application and is not intended to limit the application. Any modification, equivalent replacement, improvement, etc. within the principles of the application shall be included in the protection scope of the application.

Claims

1. An automated irrigation method for high-yield potato cultivation, characterized in that, The method includes the following steps: Obtain the soil moisture value at each outlet of each underground water pipe vertically inserted into the ground in the potato planting field at each moment along the longitudinal depth direction. A growth function for potatoes at different growth stages is constructed. Based on the magnitude of the output value of the growth function for potatoes at different growth stages in each preset period, the potato image acquisition frequency in each preset period is determined. Based on the potato image acquisition frequency in each preset period, several potato images are acquired. Each potato image corresponds to an underground water outlet and a time. Construct a neural network to identify potato maturity values ​​and obtain the trained neural network; input all potato images corresponding to each underground water outlet at the current moment into the trained neural network to obtain the potato maturity value corresponding to each underground water outlet at the current moment. Based on the potato maturity value corresponding to each underground water outlet at the current moment, determine the soil moisture requirement value corresponding to each underground water outlet at the current moment, including: constructing a potato maturity value-soil moisture requirement distribution function, inputting the potato maturity value corresponding to each underground water outlet at the current moment into the potato maturity value-soil moisture requirement distribution function, and outputting the soil moisture requirement value corresponding to each underground water outlet at the current moment. Based on the comparison between the soil moisture requirement value corresponding to each underground water outlet pipe at the current moment and the soil moisture value at each outlet of each underground water outlet pipe at the current moment, the solenoid valve switching result of each outlet of each underground water outlet pipe distributed along the longitudinal depth direction at the current moment is obtained, so as to realize drip irrigation of soil at different depths.

2. The automated irrigation method for high-yield potato cultivation according to claim 1, characterized in that, The specific steps involved in constructing the growth function of potatoes at different growth stages are as follows: The duration of potato growth in different growth stages is obtained. Using the membership function, growth functions of potato are constructed for the germination stage, seedling stage, tuber formation stage, peak tuber formation stage, and end tuber formation stage, based on the duration of potato growth in different growth stages.

3. The automated irrigation method for high-yield potato cultivation according to claim 1, characterized in that, The specific steps for determining the potato image acquisition frequency within each preset period are as follows: Input any preset period into the growth function of potatoes in different growth stages, output several probability values, and determine the sampling frequency within the any preset period based on the magnitude of the maximum probability value.

4. The automated irrigation method for high-yield potato cultivation according to claim 3, characterized in that, The specific steps for determining the sampling frequency within any preset period based on the magnitude of the maximum probability value are as follows: The product of the maximum probability value and the preset basic acquisition frequency, rounded up, is taken as the acquisition frequency within any preset period.

5. The automated irrigation method for high-yield potato cultivation according to claim 1, characterized in that, The specific steps involved in constructing the neural network for recognizing potato maturity values ​​are as follows: The neural network for identifying potato maturity values ​​includes: a dataset, the final label for each potato image, and a loss function; The dataset consists of all potato images; Based on the final label, construct the loss function.

6. The automated irrigation method for high-yield potato cultivation according to claim 5, characterized in that, The final label for each potato image consists of the growth stage classification value and maturity value corresponding to each potato image.

7. The automated irrigation method for high-yield potato cultivation according to claim 1, characterized in that, The specific steps for obtaining the potato maturity value corresponding to each underground water outlet pipe at the current moment are as follows: Each potato image is input into the trained neural network to obtain the output maturity value for each potato image; The average of the output maturity values ​​of all potato images corresponding to each underground water outlet at the current moment is taken as the potato maturity value corresponding to each underground water outlet at the current moment.

8. The automated irrigation method for high-yield potato cultivation according to claim 1, characterized in that, The specific steps for obtaining the solenoid valve switching results of each outlet of each underground water pipe at the current moment are as follows: If the soil moisture requirement value corresponding to each underground water outlet pipe at the current moment is greater than the soil moisture value corresponding to each water outlet of each underground water outlet pipe at the current moment, then the solenoid valve of each water outlet of each underground water outlet pipe will be opened. If the soil moisture requirement value corresponding to each underground water outlet pipe at the current moment is less than or equal to the soil moisture value corresponding to each water outlet of each underground water outlet pipe at the current moment, then close the solenoid valve of each water outlet of each underground water outlet pipe.

9. An automated irrigation device for high-yield potato cultivation, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the computer program is executed by the processor, it implements the steps of an automated irrigation method for high-yield potato cultivation as described in any one of claims 1-8.

Images