Fruit yield prediction system based on multi-factor regulation and control

Abstract

The invention discloses a fruit yield prediction system based on multi-factor regulation and control, which comprehensively considers various influence factors such as meteorological data, natural disaster estimation and management measures, and predicts the yield change of a litchi park in a data-driven manner. The prediction mode is not simple identification or pure data speculation, machine learning and deep learning technologies are combined, and the precision of litchi yield estimation can be effectively improved. The system is characterized in that output value data, meteorological data and extreme weather conditions of the park over the years are combined, and output prediction of the large-scale park is realized in a machine learning mode. On the basis, an unmanned aerial vehicle, an inspection robot and the like are introduced for daily inspection so as to obtain more comprehensive park information, and accurate yield prediction of each small area of the park is realized through integration of the established large model and the data. Through the integration, powerful support can be provided for more perfect orchard intelligent management of the orchard.

Description

Technical Field

[0001] The present invention relates to the field of smart agricultural technology, and more particularly to a fruit yield prediction system based on multi-factor regulation. Background Art

[0002] Lychee is an important specialty fruit in southern my country, with widespread cultivation and high economic value. However, lychee yield is influenced by a variety of complex factors, including climate, pests and diseases, soil conditions, and cultivation and management techniques. Traditional empirical and simple statistical forecasting methods often struggle to accurately predict lychee yield, hindering farmers' ability to make informed planting, management, and marketing decisions, which in turn impacts economic returns.

[0003] Currently, several crop yield prediction systems based on meteorological data, remote sensing data, and farmland IoT technologies have emerged both domestically and internationally. These systems suffer from the following main issues: First, they rely on a single data source, failing to fully consider all key factors influencing fruit yield; second, their simple prediction models struggle to process nonlinear, high-dimensional data; third, their low level of intelligence lacks recommendations for optimizing orchard management measures; and fourth, their high cost hinders widespread adoption among lychee growers.

[0004] Specifically, existing meteorological data forecasting methods primarily focus on a few key meteorological factors, such as temperature, precipitation, and sunlight, while ignoring the combined effects of humidity, wind speed, and sunshine duration. Remote sensing data analysis technology can monitor the growth of lychee trees, but data quality is unstable due to factors such as cloud cover and lighting conditions. While farmland IoT technology can collect real-time information on soil, moisture, and nutrients, the high installation and maintenance costs of the equipment make it difficult to promote across large orchards. Traditional yield forecasting methods based on statistical analysis fail to capture the complex correlations in the data, resulting in low prediction accuracy. Most importantly, the lack of automated inspections and intelligent management methods has resulted in relatively low monitoring and management efficiency for lychee growth.

[0005] Therefore, there is an urgent need for a fruit yield prediction system that can comprehensively consider multiple factors, utilize advanced technologies, and achieve accurate prediction and intelligent management to improve the overall level of orchard management. Summary of the Invention

[0006] In view of this, the present invention provides a fruit yield prediction system based on multi-factor regulation that solves at least some of the above-mentioned technical problems, which can effectively improve the accuracy of fruit yield prediction and provide strong support for the intelligent management of orchards.

[0007] To achieve the above object, the technical solution adopted by the present invention is:

[0008] The present invention provides a fruit yield prediction system based on multi-factor regulation, which includes: a data acquisition module, a data processing module and a yield prediction module, wherein:

[0009] The data acquisition module is used to obtain meteorological data, natural disaster prediction data, management measures data and historical production data of the orchard;

[0010] The data processing module is used to clean, integrate and perform feature engineering on the acquired data;

[0011] The yield prediction module is used to build a yield prediction model based on machine learning and deep learning, and uses the processed data as input to predict the fruit yield of the orchard.

[0012] In an optional embodiment, the meteorological data acquired by the data acquisition module includes: data on temperature, precipitation and light; the management measure data acquired includes: data on fertilizer application amount, irrigation frequency and pruning intensity.

[0013] In an optional embodiment, the data processing module integrates the data sources into a multi-dimensional input feature vector, and constructs the vector as follows:

[0014]

[0015] Among them, X represents the input feature vector, T represents the annual average temperature, P represents the precipitation during the growing season, L represents the duration of sunlight, and T ext represents the number of days with extreme temperatures, D represents the comprehensive quantitative value of the disaster, F represents the amount of fertilizer applied, G represents the number of irrigations, R represents the pruning intensity, and H represents the moving average of historical yield data; each component in the input feature vector is preprocessed by normalization or standardization before use.

[0016] In an optional embodiment, the yield prediction module constructs a yield prediction model using an SVR model based on the constructed input feature vector X; the constructed yield prediction model is:

[0017]

[0018] Among them, f SVR (x) represents the output predicted by the SVR model, N represents the total number of samples, and X i represents the i-th sample, represents the Lagrange multiplier, K(x i , x) represents the kernel function, and b represents the bias term.

[0019] In an optional embodiment, the system further includes: an inspection module, which is composed of a drone and / or an inspection robot and is used to inspect the orchard along a preset route and collect fruit image data.

[0020] In an optional embodiment, the system further includes an edge computing module configured to receive image data from the inspection module, perform data classification, image recognition, and data purification using an image recognition algorithm, and calculate the fruit yield in each area of the orchard using the following formula:

[0021]

[0022] Among them, E k represents the predicted value of fruit yield in region k, q k represents the farmland area in region k, n k represents the number of targets detected in region k, C kι , N kι , S kι They represent the confidence, quantity and size of the ith target in region k, respectively, and m represents the total number of regions in the orchard.

[0023] In an optional embodiment, the image recognition algorithm is a YOLO algorithm.

[0024] In an optional embodiment, the system also includes: a terminal server, which is used to receive data processed by the edge computing module, and integrate these data with the yield prediction model to construct a fusion prediction model, and use the fusion prediction model to achieve accurate prediction of fruit yield.

[0025] In an optional implementation, the fusion prediction model constructed by the terminal server includes:

[0026]

[0027] Among them, f final (x) represents the output predicted by the fusion prediction model, λ represents the fusion weight coefficient, and w k represents the weight of region k, and g(·) represents the dynamic correction function.

[0028] In an optional embodiment, the fruits predicted by the system include one or more of the following: lychee, kiwi, apple and grape.

[0029] Compared with the prior art, the present invention has at least the following beneficial technical effects:

[0030] 1. This invention comprehensively considers multiple influencing factors, including meteorological data, natural disaster forecasts, and management measures, to accurately predict yield changes in lychee orchards through a data-driven approach. The system of this invention can effectively improve the accuracy of yield estimates for lychee and other fruits.

[0031] 2. The present invention combines the historical output value data, meteorological data and extreme weather conditions of the park, and realizes the production forecast of a large area of the park through machine learning. On this basis, the system introduces drones and inspection robots for daily inspections to obtain more comprehensive park information. Drones and inspection robots collect image data according to preset routes, process them through edge computing devices, and use image recognition technology (such as YOLO algorithm) to perform data classification, image recognition and data purification. Finally, the purified data is transmitted to the terminal server, and the server will integrate these data through the established large-scale model to complete the accurate production forecast of each small block in the park. Through the above integration, the present invention can effectively improve the prediction accuracy of fruit yield in orchards, and provide strong support for the park to achieve more complete intelligent management of orchards.

[0032] Other features and advantages of the present invention will be described in the following description, and in part will become apparent from the description, or will be understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description and the accompanying drawings.

[0033] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, a brief introduction will be given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0035] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention.

[0036] Figure 1 This is a schematic diagram of the structure and workflow principle of the fruit yield prediction system based on multi-factor regulation provided by an embodiment of the present invention.

[0037] Figure 2 This is a structural diagram of a data acquisition module provided in an embodiment of the present invention.

[0038] Figure 3Schematic diagram of the data processing flow of the edge computing module provided in an embodiment of the present invention.

[0039] Figure 4 Schematic diagram of inspection by drones and inspection robots provided in an embodiment of the present invention.

[0040] Figure 5 A schematic diagram of lychee fruit identification provided by an embodiment of the present invention.

[0041] Figure 6 A schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0042] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions of the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings herein can be arranged and designed in various different configurations.

[0043] In describing the present invention, it should be noted that some processes described in this specification and accompanying drawings include multiple operations that appear in a specific order. However, it should be understood that these operations may be performed in a different order than the order in which they appear, or may be performed in parallel. Furthermore, the use of various sequence numbers is for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0044] Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0045] See also Figures 1 to 5 As shown, the embodiment of the present invention provides a fruit yield prediction system based on multi-factor regulation. The system architecture mainly includes: a data acquisition module, a data processing module, a yield prediction module (mainly a yield prediction model), an inspection module, an edge computing module and a terminal server. Data is transmitted between each module through a network connection. Among them:

[0046] The data acquisition module is responsible for collecting various relevant data and uploading it to the data processing module. The data processing module cleans, integrates, converts, and performs feature engineering on the data, providing the processed data to the yield prediction model. The yield prediction model uses this processed data to make yield predictions and sends the prediction results to the terminal server. The inspection module inspects the orchard along a pre-set route, collecting image data and uploading it to the edge computing module. The edge computing module processes the image data, identifies the fruit, and sends the identification results to the terminal server. The terminal server integrates various data for accurate predictions and intelligent management.

[0047] The system of the present invention can realize visual and controllable fruit production process, make traditional agriculture more intelligent, greatly reduce the labor burden, provide strong support for the intelligent management of orchards, and help improve the economic benefits of fruits.

[0048] Taking litchi fruit as an example, the specific implementation and working principle of the fruit yield prediction system based on multi-factor regulation of the present invention are introduced in detail below:

[0049] A fruit yield prediction system based on multi-factor regulation includes the following modules:

[0050] (1) Data acquisition module: The data acquisition module includes meteorological sensors, soil sensors, cameras and other equipment. Various sensors and equipment are connected to the data acquisition terminal through a wireless network. The data acquisition terminal uploads the collected data to the data processing module. This module collects various relevant data of the litchi park, including the following data:

[0051] (1.1) Meteorological data: There are two ways to obtain meteorological data. The first is to connect to the meteorological bureau on the day to obtain local temperature, humidity, wind speed, rainfall forecast and other meteorological data. At the same time, one to three small weather stations will be installed in the park to directly obtain the park's QC10 meteorological data. The two data are integrated to obtain a final data.

[0052] (1.2) Natural Disasters: This section collects estimated information on natural disasters such as typhoons, droughts, floods, cold waves, etc., including the probability, intensity, and impact range of the disasters. This data is primarily based on data from the local meteorological bureau. A disaster prediction model for the self-organized park will be established later, if the capability is met.

[0053] (1.3) Management measures data: records the type, timing, amount, and effectiveness of management measures such as fertilization, irrigation, pruning, and pest and disease control. This data is mainly obtained through manual recording.

[0054] (1.4) Historical production data: The production data of litchi orchards over the years were collected, including total production, average yield per unit area, and production of different varieties. The records of the orchards were used, and the data source started from 2014.

[0055] (2) Data processing module: This module is responsible for cleaning, integrating, converting and feature engineering the collected data. Specifically, it includes:

[0056] (2.1) Data cleaning: Check whether the data has outliers, missing values, duplicate values, etc., and perform corresponding processing.

[0057] (2.2) Data integration: Integrate data from different data sources into a unified format to establish a unified data warehouse.

[0058] (2.3) Data conversion: Convert raw data into a format suitable for machine learning models, such as converting text data into numerical data.

[0059] (2.4) Feature Engineering: Extract useful features from raw data, such as calculating the average, maximum, minimum, median, and extreme values of temperature, and construct new features based on domain knowledge.

[0060] (3) Yield prediction model: This module is responsible for building a litchi yield prediction model using the processed data. Specifically, it includes:

[0061] (3.1) Model selection: A variety of machine learning and deep learning algorithms can be selected, such as linear regression, support vector machine, random forest, neural network, etc. Considering that litchi yield is affected by a combination of factors, the support vector regression (SVR) model is currently used.

[0062] (3.2) Model Training: To accurately predict litchi yield, we used historical data, including the past decade's litchi yield and meteorological data, to train the model and adjust its parameters. The training process included partitioning the data into training, validation, and test sets; processing the data through feature engineering; training the SVR model on the training set (selecting the kernel function and adjusting the parameters); monitoring model performance on the validation set to avoid overfitting; and finally, evaluating the model's generalization ability on the test set. Based on the evaluation results, we continuously optimized the model until the required prediction accuracy was met. Finally, the trained model was deployed on the terminal server and regularly updated to ensure prediction accuracy.

[0063] (3.3) Model evaluation: Use test data to evaluate the model and calculate indicators such as prediction accuracy, recall rate, and F1 value.

[0064] (3.4) Model optimization: Based on the evaluation results, the model is optimized, such as adjusting the model structure, parameters, feature selection, etc.

[0065] (4) Inspection module: see Figure 4 As shown, this module consists of drones and / or inspection robots, responsible for daily inspections of litchi orchards and collecting image data. Specifically, it includes:

[0066] (4.1) Inspection route planning: Plan a reasonable inspection route based on the actual situation of the litchi orchard.

[0067] Following the following principles can meet the route planning requirements of the present invention to the greatest extent.

[0068] Power consumption control: During a patrol inspection, the power consumption of the drone or inspection vehicle shall not exceed 60%, leaving sufficient margin to deal with emergencies.

[0069] Route continuity: Inspection routes are designed to avoid large turns or reversals, maintaining continuity as much as possible to reduce unnecessary energy loss and time waste.

[0070] Coverage rate is prioritized: We strive to cover the entire litchi orchard in one round trip, ensuring that more than 80% of the litchi trees in each area can be seen.

[0071] Shooting point optimization: Set up several fixed shooting points in each area. Drones or inspection vehicles must hover or park stably after arriving at the shooting points to capture clear and stable images of the target. The setting of these shooting points requires careful consideration of shooting angles, lighting conditions, and obstructions.

[0072] (4.2) Image Data Acquisition: Drones and / or inspection robots conduct inspections along pre-set routes and collect image data using high-definition cameras. The cameras use multi-stage zoom lenses and have an image resolution of 8064x6048.

[0073] (4.3) Data upload: The collected image data is uploaded to the cloud or edge computing device in real time.

[0074] (5) Edge computing module: This module is responsible for processing the image data uploaded by the inspection module. Specifically, it includes:

[0075] (5.1) Data preprocessing: Perform preprocessing operations such as denoising and enhancement on image data to improve image quality.

[0076] (5.2) Object detection: Use object detection algorithms such as the YOLOV algorithm to identify litchi fruits in the image and count the number, size, maturity, and other information of the fruits.

[0077] (5.3) Data purification: Filter out erroneous information in the detection results, such as removing leaves that are mistakenly identified as lychee fruits.

[0078] (6) Terminal Server: This server is responsible for integrating various data to achieve accurate prediction and intelligent management. Specifically includes:

[0079] (6.1) Data integration: Integrate the data from the data acquisition module and edge computing module to establish a unified data set.

[0080] (6.2) Yield prediction: Use the yield prediction model to predict the yield of the litchi orchard and visualize the prediction results.

[0081] In a specific embodiment, preferably, the meteorological data mainly utilizes meteorological factors such as temperature, precipitation, and sunlight, etc. The management measure data mainly utilizes management measure data such as fertilization, irrigation, and pruning, etc.

[0082] In one specific embodiment, the YOLO V8 algorithm is preferably used for image recognition. The YOLO V8 model structure is as follows: the YOLO V8 model consists of an input layer (responsible for image preprocessing), a backbone network (CSPDarknet53, used to extract image features), a neck network (using FPN to further fuse features), and a detection head (using a decoupled head structure, performing localization and classification prediction separately). In addition, information related to the loss function is provided. By constructing target box regression, category classification loss, and confidence loss, the algorithm results can be optimized.

[0083] Training process: This mainly involves preprocessing the data related to the model algorithm. The gradient descent algorithm is also used to optimize and adjust the parameters related to the network structure.

[0084] Inference process: The input image is adjusted and sent to the model. The forward calculation predicts the target box position and confidence. Post-processing methods such as NMS are used to complete the result screening. Finally, the final judgment and data optimization are made on the image information.

[0085] In a specific embodiment, the specific mathematical expression involved in the prediction model is as follows:

[0086] 1. Data preprocessing and feature construction:

[0087] First, the symbols of each data source are defined and preprocessed (normalized or standardized) to ensure that data of different dimensions are comparable.

[0088] 1.1 Meteorological data includes:

[0089] Average annual temperature: T

[0090] Precipitation during growing season: P

[0091] Light duration: L

[0092] Extreme temperature days: T ext .

[0093] 1.2 Natural Disaster Forecast:

[0094] Disaster index: Set the disaster index D to represent the comprehensive quantitative value of various disasters such as typhoons, droughts, and floods. It can be specifically defined as:

[0095] D=ω1D typhoon +ω2D drought +ω3D flood ,

[0096] Among them, D typhoon 、D drought 、D flood are the probability or intensity of typhoon, drought and flood respectively, W i is the weight of each disaster.

[0097] 1.3 Management measures data include:

[0098] Fertilizer amount: F

[0099] Irrigation times: G

[0100] Pruning strength: R.

[0101] 1.4 Historical production data:

[0102] Historical production data: Y t-1 , Y t-2 ,......Y t-k To capture trends, a moving average or trend indicator can be constructed:

[0103] Moving Average:

[0104] 1.5 UAV / inspection robot image analysis results:

[0105] For each image I, multiple targets are detected by the YOLOV8 model, and their information is recorded as:

[0106] YOLO(I)={(C v , N v , S v )|v=1,2,dots,n}

[0107] Where: C v represents the target confidence score, N v Indicates the number of identified lychees, S v Represents the size of the litchi, n represents the number of objects detected in the image. At the same time, the area of the farmland represented by the image I is set to q.

[0108] Furthermore, the above data sources are integrated into a multi-dimensional input feature vector. The vector is defined as:

[0109]

[0110] In this embodiment, each component is pre-processed (such as normalization or standardization) before actual use.

[0111] 2.SVR model establishment:

[0112] In this embodiment, a support vector regression (SVR) model is established based on the constructed input feature vector X. The basic SVR prediction function is defined as:

[0113]

[0114] Where: N is the total number of training samples; X i represents the i-th sample, whose components are the same as X; is the Lagrange multiplier, obtained by solving the dual problem;

[0115] K(x i , x) is the kernel function (such as RBF kernel):

[0116] K(x i , x) = exp(-γ|x i -x| 2 )

[0117] Among them, b is the bias term.

[0118] The SVR model constructs the objective function in the original problem, which is usually expressed as:

[0119]

[0120] Where W is the weight vector, It represents minimizing the norm of the weight vector w and controlling the complexity of the model.

[0121] Satisfy the constraints:

[0122]

[0123] Where: y i is the actual litchi production (target variable); φ(x i ) is a function that maps to a high-dimensional feature space (the kernel method implies this mapping); ∈ is the tolerance error range; C is the penalty coefficient, which is used to balance the model complexity and training error; It is a slack variable that allows some samples to exceed the error range but will be penalized.

[0124] Through dual transformation, we can get the dual problem:

[0125]

[0126] The above formula satisfies the condition:

[0127] After solving, the bias b is calculated through the KKT condition of the support vector to obtain the complete prediction function f SVR (x).

[0128] 3. Image data processing and litchi size estimation:

[0129] For each image I, the YOLOV8 model is used to detect the target and obtain the target detection result: the area of farmland represented by image I is q, and the estimated size of litchi per unit area is:

[0130]

[0131] If the entire park is divided into m small areas, the image data in each small area k is processed to obtain:

[0132]

[0133] Among them, q k is the farmland area represented by region k, n k is the number of targets detected in region k, C kι , N kι , S kι are the confidence, number and size of the ith target in region k, respectively.

[0134] Furthermore, the litchi size of the entire park is the sum of the estimated values of each area:

[0135]

[0136] 4. Fusion prediction model design:

[0137] In this embodiment, in order to utilize traditional machine learning (SVR) and image data, two fusion methods are designed. Specifically:

[0138] The prediction results of the SVR model are weighted combined with the litchi size estimation obtained from the image data. The fused prediction function is:

[0139] f final (x) = f SVR (x)+λE total

[0140] Where, f SVR(x) represents the yield predicted by the SVR model; λ represents the fusion weight coefficient (determined by experiments).

[0141] The dynamic correction function g(·) is used to adjust the SVR prediction results to make them more adaptive.

[0142] The two forms are as follows:

[0143] Based on the dynamic correction of the overall image data, the image estimation value E of the entire park is used. total , then:

[0144] f final (x) = f SVR (x)·(1+λg(E total ))

[0145] Commonly used dynamic correction functions can be Sigmoid function or Tanh function:

[0146] or g(E total )=tanh(E total ).

[0147] Complex dynamic correction based on regional weighting:

[0148] After correcting each small area individually, the weighted sum is calculated and recorded as w k is the weight of region k, then:

[0149]

[0150] Among them, f final (x) represents the output predicted by the fusion prediction model.

[0151] From the description of the above embodiments, those skilled in the art will appreciate that the present invention provides a fruit yield prediction system based on multi-factor regulation. This system comprehensively considers multiple influencing factors, including meteorological data, natural disaster forecasts, and management measures, to accurately predict yield changes in litchi orchards through a data-driven approach. This prediction method goes beyond simple identification or pure data inference. By combining machine learning and deep learning techniques, the model can effectively improve the accuracy of litchi yield estimates. A key feature of this system model is that it combines historical park output data, meteorological data, and extreme weather conditions, enabling yield prediction for a large area of the park through machine learning. Furthermore, the system also incorporates drones and inspection robots for daily inspections to obtain more comprehensive park information. The drones and inspection robots collect image data along pre-set routes and upload it to the cloud in real time. The cloud data is processed by edge computing devices, using image recognition technology (such as the YOLO algorithm) for data classification, image recognition, and data purification. Finally, the purified data is transmitted to a terminal server, which integrates this data with an established large-scale model to achieve accurate yield predictions for each small block in the park.

[0152] Through this integration, the system can effectively improve the yield prediction accuracy of lychee orchards and provide strong support for more comprehensive intelligent orchard management. Compared with traditional methods, traditional fruit farmers lack foresight about harvests and rely solely on experience to make predictions. This system can minimize errors and allow for proactive preparations to address unusual situations. Furthermore, drone patrols significantly improve efficiency compared to manual labor, while also reducing labor costs.

[0153] Further, refer to Figure 6 As shown, an embodiment of the present invention also provides an electronic device for predicting fruit yield using the above system. The electronic device may include a processor 10, a memory 11, a communication bus 12 and a communication interface 13, and may also include a computer program stored in the memory 11 and executable on the processor 10.

[0154] In some embodiments, the processor 10 may be composed of an integrated circuit, such as a single packaged integrated circuit, or a plurality of packaged integrated circuits with the same or different functions, including one or more central processing units (CPUs), microprocessors, digital processing chips, graphics processors, and a combination of various control chips. The processor 10 is the control core (Control Unit) of the electronic device, connecting the various components of the entire electronic device using various interfaces and circuits, and executing programs or modules stored in the memory 11 (computer-readable storage medium) and calling data stored in the memory 11 to perform various functions of the electronic device and process data.

[0155] Examples of computer-readable storage media include read-only memory (ROM), random access programmable read-only memory (PROM), electrically erasable programmable read-only memory (EEPROM), random access memory (RAM), dynamic random access memory (DRAM), static random access memory (SRAM), flash memory, non-volatile memory, CD-ROM, DVD-ROM, Blu-ray or optical disk storage, hard disk drive (HDD), solid-state drive (SSD), card-type memory (such as a multimedia card, a secure digital (SD) card, or an ultra-fast digital (XD) card), magnetic tape, floppy disk, magneto-optical data storage device, optical data storage device, hard disk, solid-state disk, and any other device configured to store a computer program and any associated data, data files, and data structures in a non-transitory manner and provide the computer program and any associated data, data files, and data structures to a processor or computer so that the processor or computer can execute the computer program. The computer program in the above-mentioned computer-readable storage medium can be executed in an environment deployed in a computer device such as a client, a host, an agent device, a server, etc.

[0156] Those skilled in the art will appreciate that embodiments of the present invention may be provided as products such as software systems or electronic devices. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0157] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0158] The above description of the disclosed embodiments is intended to enable one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be readily apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention is not limited to the embodiments shown herein but is intended to conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A fruit yield prediction system based on multi-factor regulation, characterized in that: The system includes: data acquisition module, data processing module and output prediction module, among which: The data acquisition module is used to obtain meteorological data, natural disaster prediction data, management measures data and historical production data of the orchard; The data processing module is used to clean, integrate and perform feature engineering on the acquired data; The yield prediction module is used to build a yield prediction model based on machine learning and deep learning, and uses the processed data as input to predict the fruit yield of the orchard.

2. A fruit yield prediction system based on multi-factor regulation according to claim 1, characterized in that: The meteorological data acquired by the data acquisition module include: temperature, precipitation and light data; the management measures data acquired include: fertilizer application amount, irrigation times and pruning intensity data.

3. A fruit yield prediction system based on multi-factor regulation according to claim 2, characterized in that: The data processing module integrates the data sources into a multi-dimensional input feature vector, and constructs the vector as follows: Among them, X represents the input feature vector, T represents the annual average temperature, P represents the precipitation during the growing season, L represents the duration of sunlight, and T ext represents the number of days with extreme temperatures, D represents the comprehensive quantitative value of the disaster, F represents the amount of fertilizer applied, G represents the number of irrigations, R represents the pruning intensity, and H represents the moving average of historical yield data; each component in the input feature vector is preprocessed by normalization or standardization before use.

4. A fruit yield prediction system based on multi-factor regulation according to claim 3, characterized in that: The yield prediction module constructs a yield prediction model based on the constructed input feature vector X using the SVR model; the constructed yield prediction model is: Among them, f SVR (x) represents the output predicted by the SVR model, N represents the total number of samples, and X i represents the i-th sample, represents the Lagrange multiplier, K(x i , x) represents the kernel function, and b represents the bias term.

5. A fruit yield prediction system based on multi-factor regulation according to claim 4, characterized in that: The system also includes: an inspection module, which is composed of a drone and / or an inspection robot and is used to inspect the orchard along a preset route and collect fruit image data.

6. A fruit yield prediction system based on multi-factor regulation according to claim 5, characterized in that: The system also includes an edge computing module, which is used to receive image data from the inspection module, perform data classification, image recognition, and data purification using an image recognition algorithm, and calculate the fruit yield in each area of the orchard using the following formula: Among them, E k represents the predicted value of fruit yield in region k, q k represents the farmland area in region k, n k represents the number of targets detected in region k, C kι ,N kι ,S kι They represent the confidence, quantity and size of the ith target in region k, respectively, and m represents the total number of regions in the orchard.

7. A fruit yield prediction system based on multi-factor regulation according to claim 6, characterized in that: The image recognition algorithm is the YOLO algorithm.

8. A fruit yield prediction system based on multi-factor regulation according to claim 6, characterized in that: The system also includes: a terminal server, which is used to receive data processed by the edge computing module, integrate these data with the yield prediction model, construct a fusion prediction model, and use the fusion prediction model to achieve accurate prediction of fruit yield.

9. A fruit yield prediction system based on multi-factor regulation according to claim 8, characterized in that: The fusion prediction model constructed by the terminal server includes: Among them, f final (x) represents the output predicted by the fusion prediction model, λ represents the fusion weight coefficient, and w k represents the weight of region k, and g(·) represents the dynamic correction function.

10. The fruit yield prediction system based on multi-factor regulation according to claim 1, characterized in that: The fruits predicted by the system include one or more of the following: lychee, kiwi, apple, and grape.

Images