Grape disease prediction method and device based on Internet of Things, terminal and medium

Abstract

The invention relates to a grape disease prediction method and device based on the Internet of Things, a terminal and a medium, and belongs to the technical field of grape disease prediction.The grape disease prediction method comprises the steps that actual image information and actual environment data of grapes in a detection area are collected; according to the actual image information, determining the grape type, the growth stage and the actual surface characteristics of the current grape; determining standard surface features and standard environment data corresponding to the growth stage according to the grape variety; comparing the standard surface features with the actual surface features and comparing the standard environment data with the actual environment data; and if the surface features are not matched with the environmental data, calling a disease statistical table matched with the growth stage, screening a first predicted disease matched with the actual environmental data and the actual surface features, and sending first early warning information. The method has the beneficial effects that the disease prevention capability in the grape planting process is effectively improved, and the economic loss risk caused by diseases is reduced.

Description

Technical Field

[0001] The present application relates to the technical field of grape disease prediction, and in particular to a grape disease prediction method, device, terminal and medium based on the Internet of Things. Background Art

[0002] Grape cultivation is an important part of modern agriculture. With the continuous improvement of agricultural modernization, precision agriculture technology has gradually become one of the key factors in promoting agricultural production efficiency. By introducing advanced monitoring and prediction methods, the level of crop health management can be effectively improved, thereby reducing the incidence of pests and diseases, reducing the use of pesticides, ensuring the quality of agricultural products and food safety, while optimizing resource allocation and promoting sustainable development. At present, many regions have formed a grape planting industry, and the research on grape planting is also relatively in-depth.

[0003] However, in terms of grape disease management, traditional methods still rely on manual observation and empirical judgment. Specifically, it is usually necessary to regularly patrol the fields and observe changes in leaves, fruits and other parts with the naked eye to discover potential disease problems. In addition, diagnosis is assisted by microscopic examination of infected samples or chemical reagents to test soil composition. In recent years, some new tools and technologies have also been applied to this field, such as portable spectrometers that can be used to quickly detect changes in plant pigment content; drones equipped with multi-spectral cameras that can cover large areas for remote sensing imaging analysis, etc. Although these methods have their own characteristics and have made up for the shortcomings of relying solely on human operations to a certain extent, they still have many limitations.

[0004] The above-mentioned conventional methods generally have the following defects: they are unable to fully and accurately capture the early signs of subtle lesions, so that they cannot play a good role in disease prevention, thereby delaying the best time for prevention and control, resulting in a high risk of economic losses. Summary of the invention

[0005] In order to overcome the above technical problems, the present application provides a grape disease prediction method, device, terminal and medium based on the Internet of Things.

[0006] In the first aspect, the present application provides a grape disease prediction method based on the Internet of Things, which adopts the following technical solutions: A grape disease prediction method based on the Internet of Things, comprising: Collect actual image information of grapes in the detection area and actual environmental data of the grapes; Determining the grape type, growth stage, and actual surface features of the current grapes according to the actual image information; According to the grape variety, determining the standard surface characteristics and standard environmental data corresponding to the growth stage; Comparing the standard surface features with the actual surface features and the standard environment data with the actual environment data; If the actual surface feature does not match the standard surface feature and the actual environmental data does not match the standard environmental data, then a disease statistics table matching the growth stage is retrieved; A first predicted disease matching the actual environmental data and the actual surface features is selected from the disease statistical table, and a first warning message is sent.

[0007] By adopting the above technical solution, a grape disease prediction method based on the Internet of Things is realized. This method comprehensively collects and analyzes the actual image information and environmental data of grapes in the detection area, combines the grape type, growth stage and its corresponding standardized data for accurate comparison, and can quickly locate the potential disease type when both the surface characteristics and environmental data are abnormal, and issue early warning information in time, thereby effectively improving the disease prevention ability in the grape planting process and reducing the risk of economic losses caused by diseases.

[0008] Optionally, the step of selecting a first predicted disease matching the actual environmental data and the actual surface features from the disease statistical table includes: Determine the identification framework based on the disease statistics table; constructing a basic probability distribution based on the actual environmental data and the actual surface characteristics; Use Dempster's synthesis rule to synthesize the final synthesis probability; Calculating a belief function and a likelihood function according to the final composite probability; According to the belief function and the likelihood function, the predicted disease type is obtained.

[0009] By adopting the above technical solutions, a disease prediction method based on Dempster-Shafer evidence theory is realized, which improves the accuracy and reliability of grape disease prediction. The specific effects are as follows: 1) Determine the identification framework, clarify the basic elements and scope of disease prediction, and ensure the consistency of subsequent calculations. 2) Construct basic probability distribution, fully consider the uncertainty of actual environmental data and actual surface characteristics, and improve the data adaptability of the model. 3) Use Dempster synthesis rules to synthesize the final synthetic probability, effectively integrate multi-source information, and enhance the credibility of the prediction results. 4) Calculate belief functions and likelihood functions, further quantify the confidence level of the prediction results, and provide a scientific basis for decision-making.

[0010] Optionally, before obtaining the predicted disease type according to the belief function and the likelihood function, the method further comprises: Obtaining a difference between the belief function and the likelihood function; Determine whether the absolute value of the difference exceeds a difference threshold; If not, the predicted disease type is obtained according to the belief function and the likelihood function.

[0011] By adopting the above technical solution, the accuracy of disease prediction results can be further improved by evaluating the difference between the belief function and the likelihood function. Specifically, when the difference does not exceed the set threshold, it indicates that the prediction model has a high confidence level. At this time, the disease can be directly searched based on the final prediction value, thereby reducing the misjudgment rate and improving the prediction efficiency.

[0012] Optionally, the determining whether the difference exceeds a difference threshold further includes: If yes, the current detection point is defined as an abnormal point, and the frequency and density of collecting surface features and environmental data corresponding to the current grape variety are increased; If the number of abnormal points that appear in the same detection time exceeds the first number threshold, the detection time is defined as the abnormal detection time; If the number of occurrences of the abnormal detection time exceeds the second numerical threshold within a detection cycle, the predicted disease type with the largest number of occurrences among all abnormal points is selected as the final predicted disease type.

[0013] By adopting the above technical solution, when the difference between the belief function and the likelihood function exceeds the difference threshold, the abnormal situation can be identified in time and the current detection point can be defined as an abnormal point, thereby improving the reliability of the prediction model. At the same time, by increasing the frequency and density of collecting surface features and environmental data corresponding to the current grape variety, the accuracy and comprehensiveness of the data are further enhanced. Within the same detection time, if the number of occurrences of the abnormal point exceeds the set first number threshold, the time period is marked as abnormal detection time, which helps to focus on the period when the problem is high. If the abnormal detection time occurs frequently and exceeds the second number threshold, the final prediction value with the most occurrences among all abnormal points is statistically analyzed to ensure that the prediction result is more representative. Finally, the disease type is accurately matched according to the selected prediction value, which effectively improves the accuracy and response speed of grape disease prediction.

[0014] Optionally, after comparing the standard surface feature with the actual surface feature and the standard environment data with the actual environment data, the method further includes: If only the actual surface feature does not match the standard surface feature, determining whether the actual environment data has abnormal fluctuations within the first time window; If so, obtain the detection point whose fluctuation is greater than the standard fluctuation curve; Filter similar environmental data and obtain the average environmental data; Selecting a second predicted disease that matches the average environmental data and the actual surface features from the disease statistical table, and sending a second warning message; If not, a first predicted disease matching the actual environmental data and the actual surface features is selected from the disease statistical table, and a first warning message is sent.

[0015] By adopting the above technical solution, when the actual surface features do not match the standard surface features, the abnormal fluctuation of the actual environmental data within the first time window is further judged. If there is an abnormal fluctuation, the accuracy of the data is improved by screening similar environmental data and obtaining the average environmental data, thereby screening out a more accurate second predicted disease and sending a second warning message; if there is no abnormal fluctuation, the first predicted disease is directly screened and the first warning message is sent. This method effectively reduces the misjudgment rate and improves the accuracy and reliability of grape disease prediction.

[0016] Optionally, the step of increasing the frequency and density of collecting surface features and environmental data corresponding to the current grape variety includes: Get the absolute value of the difference and the offset value of the difference threshold ; , is the difference, is the difference threshold; According to the offset value And the preset classification adjustment rules to determine the frequency coefficient and density coefficient ; Get the base sampling frequency and baseline sampling density ; ， N is the number of initial image acquisition devices turned on, S is the area of ​​the detection area, and K is the regional complexity coefficient; ; The reference sampling frequency , the frequency coefficient and the offset value Input the sampling frequency adjustment model to obtain the adjusted new sampling frequency ; The sampling frequency adjustment model is ; According to the new sampling frequency , determine the collection interval; The benchmark sampling density , the density coefficient and the offset value Input the sampling density adjustment model to obtain the adjusted new sampling density ; The sampling density adjustment model is , is the area compensation coefficient, is the critical area, i.e. the area of ​​the entire grape growing area; According to the new sampling density , get the number of newly opened image acquisition devices ; ; According to the detection area S, the total number of sensor groups is obtained. ; , m is the redundancy coefficient, To cope with sudden failures, is the baseline acquisition density of the sensor group; According to the offset value and the total arrangement quantity , get the number of newly opened groups , .

[0017] Optionally, after comparing the standard surface feature with the actual surface feature and the standard environment data with the actual environment data, the method further includes: If only the actual environment data does not match the standard environment data, determining whether the actual environment data continues to not match within a second time window; If yes, a third predicted disease matching the actual environmental data is selected from the disease statistical table, and a third warning information is sent; If not, a first predicted disease matching the actual environmental data and the actual surface features is selected from the disease statistical table, and a first warning message is sent.

[0018] By adopting the above technical solution, when the actual environmental data does not match the standard environmental data, it is possible to further determine whether the mismatch is persistent. If the actual environmental data continues to mismatch within the second time window, a more accurate third predicted disease can be screened out from the disease statistics table and a third warning message can be sent, thereby improving the accuracy of disease prediction. If the continuous mismatch condition is not met, the next best option is to match the first predicted disease based on the actual environmental data and the actual surface features and send the first warning message, thereby ensuring the robustness and comprehensiveness of the prediction system.

[0019] In the second aspect, the present application provides a grape disease prediction device based on the Internet of Things, which adopts the following technical solutions: A grape disease prediction device based on the Internet of Things, comprising: An image acquisition module is used to acquire actual image information of grapes in the detection area; A data collection module is used to collect the actual environmental data of the grapes in the detection area; An image analysis module, for determining the grape type, growth stage and actual surface features of the current grapes according to the actual image information; A screening and matching module, for determining, according to the grape variety, the standard surface features and standard environmental data corresponding to the growth stage; A comparison processing module, used for comparing the standard surface features with the actual surface features and the standard environmental data with the actual environmental data; if the actual surface features do not match the standard surface features and the actual environmental data do not match the standard environmental data, then retrieving a disease statistics table that matches the growth stage; The screening and matching module is used to screen the first predicted disease that matches the actual environmental data and the actual surface features from the disease statistical table, and send a first warning message.

[0020] In a third aspect, the present application provides a terminal, which adopts the following technical solution: A terminal, comprising: A memory storing a grape disease prediction program based on the Internet of Things; The processor is used to execute the program stored in the memory to implement the steps of the above-mentioned grape disease prediction method based on the Internet of Things.

[0021] In a fourth aspect, the present application provides a computer-readable storage medium, which adopts the following technical solution: A computer-readable storage medium stores a computer program that can be loaded by a processor and execute the above-mentioned grape disease prediction method based on the Internet of Things.

[0022] In summary, the present application includes at least one of the following beneficial technical effects: By collecting actual image information and environmental data of grapes and comparing them with standard surface features and standard environmental data, the growth status and potential diseases of grapes can be accurately identified, which significantly improves the accuracy of disease prediction and solves the problem that traditional methods are difficult to fully capture early lesions. A multi-level early warning mechanism based on different matching situations, such as abnormal fluctuation judgment and time window analysis, further improves the reliability of disease prediction, reduces the risk of misjudgment and missed diagnosis, and ensures timely implementation of prevention and control measures; Combining the Dempster synthesis rule and disease statistics table to achieve scientific and quantitative prediction of disease types not only enhances the credibility of the prediction results, but also provides strong support for subsequent precise prevention and control. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1It is a flowchart of an implementation method of Example 1 of the present application; Figure 2 It is a flowchart of a specific step of S160; Figure 3 It is a schematic diagram of the combined probability of four pieces of evidence; Figure 4 It is a flowchart of the steps that can be executed after the difference between the belief function and the likelihood function exceeds the difference threshold; Figure 5 It is a flowchart for improving the frequency and density of acquisition; Figure 6 It is an implementation method of the steps that can be performed after S140; Figure 7 is another implementation of the steps that can be performed after S140; Figure 8 It is a structural block diagram of an implementation method of the first embodiment of the system of the present application. DETAILED DESCRIPTION

[0024] To make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the following will be combined with the appended drawings of the embodiments of the present invention. Figure 1 -Attached Figure 8 , the technical solutions in the embodiments of the present invention are clearly and completely described. Obviously, the described embodiments are part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0025] The first embodiment of the present application discloses a grape disease prediction method based on the Internet of Things. Figure 1 As an implementation of the grape disease prediction method, the grape disease prediction method may include S110-S160: S110, collecting actual image information of grapes in the detection area and actual environmental data of the grapes; S120, determining the grape type, growth stage, and actual surface features of the current grapes according to the actual image information; S130, determining standard surface characteristics and standard environmental data corresponding to the growth stage according to the grape variety; S140, comparing the standard surface features with the actual surface features and the standard environment data with the actual environment data; S150, if the actual surface characteristics do not match the standard surface characteristics and the actual environmental data do not match the standard environmental data, then retrieve a disease statistics table that matches the growth stage; S160, selecting a first predicted disease that matches the actual environmental data and the actual surface features from the disease statistical table, and sending a first warning message.

[0026] Specifically, the surface features of grapes include the area of ​​leaf spots and the degree of fruit discoloration; environmental data include temperature data and humidity data. Taking grapes grown in a greenhouse as an example, image acquisition devices such as high-resolution digital cameras or cameras are installed in the greenhouse. The area covered by the image acquisition device is defined as the detection area, which can be understood as a detection area containing a number of image acquisition devices. In addition, a number of temperature sensors and humidity sensors are also installed in the detection area. The temperature sensor and humidity sensor can be installed in a fixed position or in a mobile position to collect data at different positions through movement.

[0027] The collected grape images are classified and the grape types are identified using a deep learning-based image recognition model (such as YOLOv4 or YOLOv5s). The deep learning-based image recognition model is trained based on a large amount of historical data of grape types. The growth stage of the grapes is determined by analyzing the color, shape, and texture characteristics of the grape leaves. The spot area and the degree of fruit discoloration on the grape surface are extracted using image processing algorithms (such as edge detection and color histogram analysis); the image processing algorithm is a conventional recognition algorithm.

[0028] Based on historical data and expert experience, a library of standard surface features of different grape varieties at different growth stages is established, including leaf spot area, fruit discoloration degree, etc. Standard environmental data (temperature, humidity range) corresponding to different growth stages are collected and associated with standard surface features to generate a grape comparison library.

[0029] Use image processing algorithms (such as convolutional neural networks) to compare the actual surface features collected with the standard surface features and calculate the similarity. Since one generation stage may correspond to multiple standard surface features, the actual surface features are compared with each standard surface feature. If the similarity after each comparison is below the similarity threshold, it is determined that the actual surface features do not match the standard surface features.

[0030] One growth stage corresponds to multiple standard temperature ranges and multiple standard humidity ranges; if the actual temperature is not within the multiple standard temperature ranges and the actual humidity is not within the multiple standard humidity ranges, it is determined that the actual environmental data does not match the standard environmental data.

[0031] The growth stages of grapes usually include budding, new shoot growth, flowering, berry growth, berry ripening, leaf fall and dormancy. Different types of grapes have different sensitivities to environmental conditions and diseases at different growth stages. Common grape diseases include downy mildew, powdery mildew, anthracnose, gray mold, etc. The occurrence of these diseases can be obtained through field surveys, laboratory tests, and historical grape disease occurrence data, and the grape type, growth stage, surface characteristics, environmental data, disease type, disease severity, etc. are recorded. A disease value range is assigned to different disease severity levels of disease types. For example, for downy mildew, the range of level 0 disease values ​​can be set to 0-5; level 1 is 6-20; level 2 is 21-40; level 3 is 41-70; level 4 is 71-100; these data are associated into a piece of disease information.

[0032] According to the grape variety and growth stage, the disease statistics table matching the current growth stage is retrieved from the disease database. The disease statistics table refers to multiple sets of disease information corresponding to the current grape variety at the current growth stage. The disease database records multiple sets of disease information corresponding to each growth stage corresponding to each grape variety.

[0033] After matching the first predicted disease according to the actual surface features and the actual environmental data, a first warning information will be sent, for example, the first warning information will be sent to a background monitoring room.

[0034] Reference Figure 2 Further, the step of selecting the first predicted disease matching the actual environmental data and the actual surface features from the disease statistical table may specifically include S161-S165: S161, determine the disease identification framework based on the disease statistics table; S162, constructing a basic probability distribution based on actual environmental data and actual surface features; S163, using Dempster's synthesis rule to synthesize the final synthesis probability; S164, calculating a belief function and a likelihood function according to the final composite probability; S165, obtaining a predicted disease type according to the belief function and the likelihood function.

[0035] Specifically, assuming that the disease statistics table includes three types of downy mildew, powdery mildew and anthracnose, the disease identification framework can be ; Among them, A=downy mildew, B=powdery mildew, C=anthracnose.

[0036] Reference to the composite probability of four pieces of evidence Figure 3 shown.

[0037] Assume that 35% of the discolored area may support B and C, 25% of the leaf spot coverage may support B and C, high temperature of 32°C may support C, and high humidity of 85% may support C.

[0038] Step 1: Fusion of the first two pieces of evidence ( and ) Calculate the conflict coefficient ; Synthesis probability ( ):

[0039] Step 2: Fusion results Evidence with temperature

[0040] Calculate the conflict coefficient ； Synthesis probability ( ):

[0041] Step 3: Fusion results With humidity evidence

[0042] Calculate the conflict coefficient ； The final synthesis probability ( ):

[0043] The belief function (Bel) is , the likelihood function (Pls) is , that is, determine the disease type as anthrax; then determine the severity based on the deviation value between the actual environmental data and the standard environmental data and the deviation value between the actual surface features and the standard surface features; for example, if the deviation values ​​of both are greater than the maximum deviation threshold, the severity is level 4; if only the deviation value corresponding to the environmental data is greater than the maximum deviation threshold, the severity is level 3; if only the deviation value of the surface features is greater than the maximum deviation threshold, the severity is level 2; if any deviation value of the environmental data and the surface features is greater than the minimum deviation threshold and less than the maximum deviation threshold, the severity is level 1; if the deviation values ​​of both the environmental data and the surface features are less than the minimum deviation threshold, the severity is level 0.

[0044] Reference Figure 4Further, after S164 and before S165, the difference between the belief function and the likelihood function may be obtained first, and then it is determined whether the absolute value of the difference exceeds the difference threshold. If not, S165 is executed; if so, S201-S203 may be executed: S201, defining the current detection point as an abnormal point, and increasing the frequency and density of collecting surface features and environmental data corresponding to the current grape variety; S202, if the number of times an abnormal point appears within the same detection time exceeds the first number threshold, the detection time is defined as the abnormal detection time; S203, if the number of occurrences of abnormal detection time exceeds the second number threshold within a detection cycle, the predicted disease type with the largest number of occurrences among all abnormal points is selected as the final predicted disease type.

[0045] Specifically, the current detection point refers to the time point when the current grape surface features and environmental data are collected. If the absolute value of the difference between the belief function and the likelihood function corresponding to the current collection time point exceeds the preset difference threshold, the current collection time point can be defined as an abnormal point, and the collection frequency and density of the surface features and environmental data corresponding to the current grape type can be increased.

[0046] Reference Figure 5 , the steps of increasing the frequency and density of collecting surface features and environmental data may include S301-S309: S301, obtaining the absolute value of the difference and the offset value of the difference threshold ; , is the difference, is the difference threshold; S302, according to the offset value And the preset classification adjustment rules to determine the frequency coefficient and density coefficient ; S303, obtaining a reference sampling frequency and baseline sampling density ; ， N is the number of initial image acquisition devices turned on, S is the area of ​​the detection area, and K is the regional complexity coefficient; ; S304, the reference sampling frequency , frequency coefficient and offset value Input the sampling frequency adjustment model to obtain the adjusted new sampling frequency ; The sampling frequency adjustment model is ; S305, according to the new sampling frequency , determine the collection interval; S306, the baseline sampling density , density coefficient and offset value Input the sampling density adjustment model to obtain the adjusted new sampling density ; The sampling density adjustment model is , is the area compensation coefficient, is the critical area, i.e. the area of ​​the entire grape growing area; S307, according to the new sampling density , get the number of newly opened image acquisition devices ; ; S308, according to the detection area S, obtain the total number of sensor groups arranged ; , m is the redundancy coefficient, To cope with sudden failures, is the baseline acquisition density of the sensor group; S309, according to the offset value and total number of arrangements , get the number of newly opened groups , .

[0047] For example, the classification adjustment rules are: When , it is a slight offset, and the corresponding frequency coefficient =0.5, density coefficient =0.3; When , it is a moderate offset, and the corresponding frequency coefficient =1.2, density coefficient =0.8; When , it is a severe offset, and the corresponding frequency coefficient =2.0, density coefficient =1.5.

[0048] If a certain difference is 0.35, the difference threshold is 0.2, then =0.75, which is a moderate shift, so the frequency coefficient is determined =1.2, density coefficient =0.8.

[0049] If the number of image acquisition devices turned on initially is 20, and the detection area S is 80 , the total area of ​​grape growing areas is 100 , is 0.3; then ,but ,but Therefore, 38 new image acquisition devices need to be opened on the basis of the original ones, that is, a total of 58.

[0050] , ; then the collection interval can be , and then rounded up to 3, so the collection interval can be once every 3 hours. That is, the sensor group also collects once every 3 hours.

[0051] For the sensor group (including temperature sensor and humidity sensor): If , ,but Group. ,but , you need to open 201 new sensor groups based on the original ones.

[0052] If environmental data and surface features are collected every 3 hours; if the current image acquisition device is turned on 58, the number of sensor groups is 300; assuming that according to historical detection data, one image acquisition device can cover the monitoring range of 6 sensor components; when the difference corresponding to the 6 environmental data and the surface features collected by the camera exceeds the difference threshold, it means that the detection point is an abnormal point; assuming that the first count threshold is 40, if there are 45 abnormal points, exceeding the first count threshold, the moment is defined as the abnormal detection moment; if within a detection cycle, for example, within 21 hours, the number of abnormal detection moments occurs 15 times, exceeding the second count threshold of 12 times, then the disease type with the most occurrences among all abnormal points can be selected as the final predicted disease type.

[0053] In addition, if the number of times the abnormal point appears within the same detection time does not exceed the first number threshold, S165 can be directly executed.

[0054] Reference Figure 6 , further, after S140, if only the actual surface features do not match the standard surface features, S401-S405 may be executed: S401, determining whether there is abnormal fluctuation in the actual environment data within the first time window; S402, if yes, obtaining the detection point whose fluctuation is greater than the standard fluctuation curve; S403, screening similar environmental data and obtaining average environmental data; S404, selecting a second predicted disease that matches the average environmental data and the actual surface characteristics from the disease statistical table, and sending a second warning information; S405: If not, select the first predicted disease that matches the actual environmental data and the actual surface features from the disease statistical table, and send a first warning message.

[0055] Specifically, the first time window representation starts from the current detection time, and the actual environmental data is monitored every half an hour for one day. The actual monitoring curve is constructed according to the monitoring data and the monitoring time, and the actual monitoring curve is compared with the standard fluctuation curve. The standard fluctuation curve represents the monitoring curve of historical environmental data matching the grape type and growth stage within one day under normal circumstances. When comparing, the detection points greater than the standard fluctuation curve are screened out, and similar environmental data are screened, and the average environmental data is obtained; similar environmental data refers to the difference between the actual environmental data is less than the difference threshold.

[0056] The execution logic of steps S304 and S305 is the same as the execution logic of S161 to S165.

[0057] Reference Figure 7 Further, after S140, if only the actual environment data does not match the standard environment data, S501-S503 may be executed: S501, determining whether the actual environment data continues to be mismatched within the second time window; S502: If yes, select a third predicted disease that matches the actual environmental data from the disease statistics table, and send a third warning message; S503: If not, select the first predicted disease that matches the actual environmental data and the actual surface features from the disease statistical table, and send a first warning message.

[0058] Specifically, if only the actual environmental data does not match the standard environmental data, it means that there is no problem with the actual surface features, so only the actual environmental data needs to be considered. The second time window is characterized by monitoring the environmental data once every 1 hour within 12 hours. If the number of abnormal environmental data is greater than 1 / 4, it means that the actual environmental data continues to mismatch within the second time window; the abnormality refers to environmental data that is greater than the standard environmental data, and the difference between the monitored environmental data and the actual environmental data is greater than the difference threshold.

[0059] The execution logic of S502 and S503 is the same as the execution logic of S161 to S165.

[0060] Based on the above method embodiment, the second embodiment of the present application discloses a grape disease prediction device based on the Internet of Things. Figure 8As an embodiment of the grape disease prediction device, the grape disease prediction device may include: An image acquisition module is used to acquire actual image information of grapes in the detection area; A data collection module is used to collect the actual environmental data of the grapes in the detection area; An image analysis module is used to determine the grape variety, growth stage and actual surface characteristics of the current grapes based on actual image information; A screening and matching module is used to determine the standard surface features and standard environmental data corresponding to the growth stage according to the grape variety; A comparison processing module is used to compare the standard surface features with the actual surface features and the standard environmental data with the actual environmental data; if the actual surface features do not match the standard surface features and the actual environmental data do not match the standard environmental data, then a disease statistics table matching the growth stage is retrieved; The screening and matching module is used to screen the first predicted disease that matches the actual environmental data and the actual surface features from the disease statistics table, and send the first warning information.

[0061] The modules of the grape disease prediction device based on the Internet of Things correspond one to one with the grape disease prediction method based on the Internet of Things, and will not be elaborated here.

[0062] The third embodiment of the present application provides a terminal. As an implementation of the terminal, the terminal may include: a memory and a processor; wherein: The memory is used to store the above-mentioned grape disease prediction program based on the Internet of Things; The processor is used to execute the program stored in the memory to implement the steps of the above-mentioned grape disease prediction method based on the Internet of Things.

[0063] The memory may be connected to the processor via a communication bus, and the communication bus may be an address bus, a data bus, a control bus, etc.

[0064] In addition, the memory may include a random access memory (RAM) and may also include a non-volatile memory (NVM), such as at least one disk storage.

[0065] And the processor can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc.

[0066] The fourth embodiment of the present application provides a computer-readable storage medium storing a computer program that can be loaded and executed by a processor to perform the above-mentioned grape disease prediction method based on the Internet of Things.

[0067] The computer-readable storage medium can be any available medium accessible by a computer or a data storage device such as a server or a data center integrating one or more available media. Among them, the available medium can be a magnetic medium (for example, a floppy disk, a hard disk, a magnetic tape), an optical medium (for example, a DVD), or a semiconductor medium (for example, a solid-state drive), etc.

[0068] The above are all the preferred embodiments of the present application, which do not limit the protection scope of the present application in sequence. Any feature disclosed in this specification (including the abstract and the drawings), unless specifically described, can be replaced by other equivalent or similar-purpose alternative features. That is, unless specifically described, each feature is only an example of a series of equivalent or similar features.

Claims

1. A grape disease prediction method based on the Internet of Things, characterized in that: include: Collect actual image information of grapes in the detection area and actual environmental data of the grapes; Determining the grape type, growth stage, and actual surface features of the current grapes according to the actual image information; According to the grape variety, determining the standard surface characteristics and standard environmental data corresponding to the growth stage; Comparing the standard surface features with the actual surface features and the standard environment data with the actual environment data; If the actual surface feature does not match the standard surface feature and the actual environmental data does not match the standard environmental data, then a disease statistics table matching the growth stage is retrieved; A first predicted disease matching the actual environmental data and the actual surface features is selected from the disease statistical table, and a first warning message is sent.

2. The grape disease prediction method based on the Internet of Things according to claim 1, characterized in that: The step of selecting a first predicted disease matching the actual environmental data and the actual surface features from the disease statistical table comprises: Determine the identification framework based on the disease statistics table; constructing a basic probability distribution based on the actual environmental data and the actual surface characteristics; Use Dempster's synthesis rule to synthesize the final synthesis probability; Calculating a belief function and a likelihood function according to the final composite probability; According to the belief function and the likelihood function, the predicted disease type is obtained.

3. The grape disease prediction method based on the Internet of Things according to claim 2, characterized in that: Before obtaining the predicted disease type according to the belief function and the likelihood function, the method includes: Obtaining a difference between the belief function and the likelihood function; Determine whether the absolute value of the difference exceeds a difference threshold; If not, the predicted disease type is obtained according to the belief function and the likelihood function.

4. The method for predicting grape diseases based on the Internet of Things according to claim 3, characterized in that: The determining whether the absolute value of the difference exceeds the difference threshold further includes: If yes, the current detection point is defined as an abnormal point, and the frequency and density of collecting surface features and environmental data corresponding to the current grape variety are increased; If the number of abnormal points that appear in the same detection time exceeds the first number threshold, the detection time is defined as the abnormal detection time; If the number of occurrences of the abnormal detection time exceeds the second numerical threshold within a detection cycle, the predicted disease type with the largest number of occurrences among all abnormal points is selected as the final predicted disease type.

5. The method for predicting grape diseases based on the Internet of Things according to claim 1, characterized in that: After comparing the standard surface features with the actual surface features and the standard environment data with the actual environment data, the method further includes: If only the actual surface feature does not match the standard surface feature, determining whether the actual environment data has abnormal fluctuations within the first time window; If so, obtain the detection point whose fluctuation is greater than the standard fluctuation curve; Filter similar environmental data and obtain the average environmental data; Selecting a second predicted disease that matches the average environmental data and the actual surface features from the disease statistical table, and sending a second warning message; If not, a first predicted disease matching the actual environmental data and the actual surface features is selected from the disease statistical table, and a first warning message is sent.

6. The method for predicting grape diseases based on the Internet of Things according to claim 4, characterized in that: The step of increasing the frequency and density of collecting surface features and environmental data corresponding to the current grape variety comprises: Get the absolute value of the difference and the offset value of the difference threshold ; , is the difference, is the difference threshold; According to the offset value And the preset classification adjustment rules to determine the frequency coefficient and density coefficient ; Get the base sampling frequency and baseline sampling density ; ， N is the number of initial image acquisition devices turned on, S is the area of ​​the detection area, and K is the regional complexity coefficient; ; The reference sampling frequency , the frequency coefficient and the offset value Input the sampling frequency adjustment model to obtain the adjusted new sampling frequency ; The sampling frequency adjustment model is ; According to the new sampling frequency , determine the collection interval; The benchmark sampling density , the density coefficient and the offset value Input the sampling density adjustment model to obtain the adjusted new sampling density ; The sampling density adjustment model is , is the area compensation coefficient, is the critical area, i.e. the area of ​​the entire grape growing area; According to the new sampling density , get the number of newly opened image acquisition devices ; ; According to the detection area S, the total number of sensor groups is obtained. ; , m is the redundancy coefficient, To cope with sudden failures, is the baseline acquisition density of the sensor group; According to the offset value and the total arrangement quantity , get the number of newly opened groups , .

7. The method for predicting grape diseases based on the Internet of Things according to claim 1, characterized in that: After comparing the standard surface features with the actual surface features and the standard environment data with the actual environment data, the method further includes: If only the actual environment data does not match the standard environment data, determining whether the actual environment data continues to not match within a second time window; If yes, a third predicted disease matching the actual environmental data is selected from the disease statistical table, and a third warning information is sent; If not, a first predicted disease matching the actual environmental data and the actual surface features is selected from the disease statistical table, and a first warning message is sent.

8. A grape disease prediction device based on the Internet of Things, characterized in that: include: An image acquisition module is used to acquire actual image information of grapes in the detection area; A data collection module is used to collect the actual environmental data of the grapes in the detection area; An image analysis module, for determining the grape type, growth stage and actual surface features of the current grapes according to the actual image information; A screening and matching module, for determining, according to the grape variety, the standard surface features and standard environmental data corresponding to the growth stage; A comparison processing module, used for comparing the standard surface features with the actual surface features and the standard environmental data with the actual environmental data; if the actual surface features do not match the standard surface features and the actual environmental data do not match the standard environmental data, then retrieving a disease statistics table that matches the growth stage; The screening and matching module is used to screen the first predicted disease that matches the actual environmental data and the actual surface features from the disease statistical table, and send a first warning message.

9. A terminal, characterized in that: include: A memory storing a grape disease prediction program based on the Internet of Things; A processor is used to execute the program stored in the memory to implement the steps of the grape disease prediction method based on the Internet of Things as claimed in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: A computer program is stored which can be loaded by a processor and executes the grape disease prediction method based on the Internet of Things as claimed in any one of claims 1 to 7.

Images