Blueberry yield prediction method, system and equipment
The root system data is collected through the sensor network and combined with meteorological data, and the blueberry yield prediction model is constructed using the support vector machine algorithm, which solves the problem of inaccurate blueberry yield prediction in the existing technology, and achieves high-precision and real-time yield prediction, supporting agricultural production. Sustainable development.
Patent Information
- Application Number
- CN202411347344.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-26
- Publication Date
- 2024-11-12
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The existing technology is difficult to comprehensively consider the health status and environmental factors of the blueberry root system, and the lack of real-time data processing capabilities, resulting in inaccurate prediction of blueberry yields, especially under rapidly changing environmental conditions.
Root system data is collected by installing an underground sensor network, combining meteorological and soil data, the root health index is calculated using the weighted average method, and a blueberry yield prediction model is constructed using the support vector machine algorithm to make real-time predictions.
Accurate prediction of blueberry yields is achieved, the accuracy and reliability of predictions is improved, and agricultural management strategies can be adjusted in a timely manner, resource waste is reduced, and crop yield and quality are improved.
Smart Images

Figure CN118940914A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of agricultural technology, and in particular to a blueberry yield prediction method, system and equipment. Background Art
[0002] Blueberry is a nutritious and widely popular berry grown in many parts of the world. However, the yield of blueberries is affected by a variety of environmental factors, including soil conditions, climate change, irrigation systems and the health of the plants themselves. Although existing agricultural technologies have been able to predict crop yields through traditional methods, such as relying on historical yield data or through simple meteorological condition models, these methods often ignore the health of crop roots and their complex interactions with environmental factors, which is crucial for accurate yield prediction. In addition, another limitation of existing prediction methods is the lack of real-time data processing capabilities, which makes it impossible to dynamically adjust the prediction model to adapt to immediate changes in environmental conditions.
[0003] Current technologies lack a systematic solution that can comprehensively consider root health data and environmental factors and conduct real-time data analysis. In addition, most existing methods fail to provide a dynamic prediction model, which is particularly insufficient under rapidly changing environmental conditions and may lead to large deviations between the predicted results and actual yields.
[0004] Therefore, there is an urgent need to develop a blueberry yield prediction method, system and device that can utilize real-time data, root health index and environmental data in combination. Summary of the invention
[0005] Based on the above objectives, the present invention provides a blueberry yield prediction method, system and device.
[0006] A blueberry yield prediction method comprises the following steps:
[0007] S1: A network of underground sensors installed around the blueberry roots collects real-time root-related data, including soil moisture, temperature, oxygen content, root conductivity, and soil microbial activity;
[0008] S2: Clean the root data collected by S1, remove abnormal or invalid data, and uniformly process different types of data through standardized methods to ensure that the data have the same dimension;
[0009] S3: Based on the data preprocessed by S2, the root health index is calculated using the weighted average method. The root health index is used to comprehensively reflect the effects of multiple factors such as soil moisture, oxygen content, electrical conductivity, and soil microbial activity on root health.
[0010] S4: Fusion of the root health index with environmental data, including meteorological data, soil nutrient content, and irrigation water quality, using a time alignment algorithm to ensure the relevance of different data sources at the same time point, and generating a comprehensive feature vector;
[0011] S5: Based on the comprehensive feature vector generated by S4, a blueberry yield prediction model was constructed using the support vector machine algorithm to predict the yield of blueberries under different root health conditions;
[0012] S6: Use historical blueberry yield data to train the blueberry yield prediction model constructed in S5, and optimize the model parameters through cross-validation method;
[0013] S7: Based on the real-time collected root health data and environmental data, the trained blueberry yield prediction model is used to make real-time predictions of blueberry yields and generate a yield prediction report.
[0014] Optionally, the S1 specifically includes:
[0015] S11: deploying a sensor network around the root system in the blueberry planting area, wherein the sensor network consists of a soil moisture sensor, a temperature sensor, an oxygen sensor, a root conductivity sensor, and a soil microbial activity detection sensor;
[0016] S12: Detect the moisture content changes in each layer of the soil through the soil moisture sensor and collect soil moisture content data;
[0017] S13: monitor the temperature of the soil around the blueberry root system in real time through a temperature sensor and collect soil temperature data;
[0018] S14: detecting the oxygen content in the soil around the root system through an oxygen sensor and collecting oxygen content data;
[0019] S15: collecting the conductivity data of the root system in real time through the root conductivity sensor;
[0020] S16: Monitor the activity of microorganisms in the soil through soil microbial activity detection sensors and record microbial activity data.
[0021] Optionally, the S2 specifically includes:
[0022] S21: Perform a preliminary analysis on the root system data collected in S1, and identify abnormal data through threshold detection method and statistical analysis method. The threshold detection method is used to detect the data of each type of sensor according to a preset reasonable data range, and determine the data exceeding the preset reasonable data range as abnormal data; the statistical analysis method will calculate the mean and standard deviation of each type of data, and eliminate the abnormal data exceeding the range of mean ± 3 times standard deviation;
[0023] S22: Repair abnormal data and missing data, including interpolation method and historical mean method; the interpolation method performs linear interpolation based on data of adjacent time points, and the historical mean method uses the historical mean of the same time period to fill in the missing data;
[0024] S23: Standardize different types of data to ensure that different data have the same dimension. The standardization process uses the Z-score standardization method to adjust the mean of each type of data to 0 and the standard deviation to 1 to eliminate the dimensional differences between different data types.
[0025] S24: After data processing is completed, all data sets are output and compared with historical data to verify the rationality of the data. By comparing the changes in data distribution, it is ensured that the processed data does not contain abnormal points.
[0026] Optionally, the S3 specifically includes:
[0027] S31: Select characteristic parameters from the data preprocessed in S2, the characteristic parameters including soil moisture, soil temperature, oxygen content, root conductivity and soil microbial activity; and assign weights according to the degree of influence of each characteristic parameter on root health; set the soil moisture weight to w1, the soil temperature weight to w2, the oxygen content weight to w3, the root conductivity weight to w4 and the soil microbial activity weight to w5, and all weights meet the following constraint: w1+w2+w3+w4+w5=1;
[0028] S32: normalize each characteristic parameter to eliminate the magnitude differences of different parameters. The normalization adopts the minimum-maximum normalization method to adjust the value of each parameter to the range of [0, 1].
[0029] S33: Calculate the root health index by weighted average method. The root health index is used to evaluate the overall health of the root system. Specifically, the normalized characteristic parameters are multiplied by the corresponding weights, and the products are added to obtain the root health index H. index , the formula is:
[0030] H index =w1·p1′+w2·p2′+w3·p3′+w4·p4′+w5·p5′, where H index is the calculated root health index, and p1′, p2′, p3′, p4′, and p5′ are the normalized soil moisture, soil temperature, oxygen content, root conductivity, and soil microbial activity data, respectively.
[0031] Optionally, the S4 specifically includes:
[0032] S41: Acquire meteorological data, soil nutrient content data and irrigation water quality data from meteorological stations and sensors, wherein the meteorological data includes parameters of temperature, humidity, rainfall and wind speed, the soil nutrient data includes the content of nutrient elements such as nitrogen, phosphorus and potassium, and the irrigation water quality data includes pH value, salt content and concentration of harmful substances in water; and standardize the acquired environmental data to ensure that they have the same dimension as the root health index;
[0033] S42: based on the timestamps collected from different data sources, the root health index and the environmental data are time-aligned. The time alignment algorithm synchronizes the data collected at different frequencies to the same time point through interpolation or linear time interpolation to ensure that the root health index and the environmental data are consistent within the same time interval. If the data at a certain time point is missing, the interpolation method is used to supplement the data based on the data at adjacent time points.
[0034] S43: The time-aligned root health index is fused with the standardized environmental data to generate a comprehensive feature vector V.
[0035] Optionally, the S5 specifically includes:
[0036] S51: receiving the comprehensive feature vector V generated by S4, and using the feature vector as input feature data of the model;
[0037] S52: A blueberry yield prediction model is constructed using a support vector machine regression algorithm. The support vector machine regression algorithm achieves accurate prediction by finding a suitable decision boundary and maximizing the distance between data points and the boundary. The support vector machine regression model formula is: f(V)=w·V+b, wherein f(V) is the predicted value of blueberry yield, w is the weight vector of the model, b is the bias term, and V is the feature vector.
[0038] Optionally, the S6 specifically includes:
[0039] S61: Obtain blueberry production data for many years from historical records as the target variable Y of the training data set, and use the comprehensive feature vector V generated in S5 as input data; the training data set is specifically represented as: {(V1, Y1), (V2, Y2), ..., (V m , Y m )}, where V i is the feature vector at the i-th time point, Y i is the actual blueberry yield at the i-th time point, and m is the number of samples in the data set;
[0040] S62: Performing model training on the training data set, the goal is to minimize the error between the predicted value f(V) output by the model and the actual output value Y by adjusting the weight vector w and the bias term b of the model; the model training process also includes using a square error loss function to measure the difference between the predicted value and the actual value;
[0041] S63: In order to improve the generalization ability of the model and avoid overfitting, the cross-validation method is used to optimize the hyperparameters of the model, wherein the hyperparameters include the regularization parameter C and the loss threshold ∈. The cross-validation method specifically divides the training data set into k subsets, selects one subset as the validation set each time, and uses the remaining k-1 subsets for training the model; the loop is executed k times, the prediction error of each validation set is calculated, and finally the parameter combination that minimizes the error is selected;
[0042] S64: According to the cross-validation results, the model's hyperparameters C and ∈ are adjusted, and the parameter combination that minimizes the validation error is selected for final model training. The optimized blueberry yield prediction model expression is: f(V) = w * ·V+b * , where w * is the optimized weight vector, b * is the optimized bias term, and V is the comprehensive feature vector.
[0043] Optionally, the S7 specifically includes:
[0044] S71: Generate a real-time feature vector, using the standardized data structure in S2 to combine the real-time collected root health data and environmental data into a real-time comprehensive feature vector V 实时 ;
[0045] S72: Using the blueberry yield prediction model trained in S6, the real-time comprehensive feature vector V 实时 Input the model to generate real-time blueberry yield forecast value V 实时 ;
[0046] S73: Based on the real-time output forecast value V 实时 Generate a forecast report that includes real-time blueberry yield forecast data, various root health indicators, real-time environmental data, and future trend analysis of blueberry yield.
[0047] A blueberry yield prediction system, used to implement the above-mentioned blueberry yield prediction method, includes the following modules:
[0048] Data acquisition module: used to collect root health data and environmental data in real time through a sensor network installed in the blueberry planting area. The root health data includes soil moisture, soil temperature, oxygen content, root conductivity and soil microbial activity. The environmental data includes meteorological data, soil nutrient content and irrigation water quality.
[0049] Data preprocessing module: used to clean, standardize and normalize the data collected by the data acquisition module;
[0050] Feature extraction and fusion module: used to extract feature parameters from the pre-processed root health data and environmental data, and generate the root health index through the weighted average method; the feature extraction and fusion module also includes fusing the root health index with the standardized environmental data to generate a comprehensive feature vector;
[0051] Model building and training module: Build a blueberry yield prediction model based on the support vector machine regression algorithm, and use historical blueberry yield data to train and optimize the model;
[0052] Real-time prediction module: Based on the real-time comprehensive feature vector generated by the feature extraction and fusion module, the trained blueberry yield prediction model is used to perform real-time prediction of blueberry yield;
[0053] Report generation module: used to generate a blueberry yield forecast report based on the blueberry yield forecast results output by the real-time forecast module. The forecast report specifically includes real-time predicted yield value, root health status, environmental data and future blueberry yield trend analysis.
[0054] A computer device comprises a processor and a memory, wherein a computer program is stored in the memory, and when the computer program is run by the processor, the above-mentioned blueberry yield prediction method is executed.
[0055] Beneficial effects of the present invention:
[0056] The present invention provides a method for accurately predicting blueberry yield by real-time collection and comprehensive analysis of root health data and environmental data. The method can not only reflect the health status of blueberry roots in real time, but also combine environmental factors such as meteorology, soil and irrigation water quality to ensure that the prediction model can accurately capture the key variables affecting blueberry yield. This highly integrated data analysis method can greatly improve the accuracy and reliability of yield prediction, thereby providing agricultural producers with a powerful decision support tool.
[0057] The present invention optimizes the yield prediction model through the support vector machine algorithm, which can effectively process multi-dimensional feature vectors and adapt to complex nonlinear relationships, ensuring that high prediction accuracy can be maintained under changing environmental conditions. This real-time prediction capability enables agricultural producers to adjust management strategies in a timely manner, such as irrigation, fertilization, and disease control, thereby reducing resource waste while improving crop yield and quality and achieving sustainable development of agricultural production. BRIEF DESCRIPTION OF THE DRAWINGS
[0058] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only for the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0059] Figure 1 A schematic diagram of a method for predicting blueberry yield according to an embodiment of the present invention;
[0060] Figure 2 Schematic diagram of a blueberry yield prediction system according to an embodiment of the present invention. DETAILED DESCRIPTION
[0061] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. At the same time, it is explained here that in order to make the embodiments more detailed, the following embodiments are the best and preferred embodiments, and those skilled in the art may also adopt other alternatives to implement some known technologies; and the accompanying drawings are only for more specific description of the embodiments, and are not intended to specifically limit the present invention.
[0062] It should be noted that the references to "one embodiment", "an embodiment", "an exemplary embodiment", "some embodiments" and the like in the specification indicate that the embodiments described may include specific features, structures or characteristics, but not every embodiment may include the specific features, structures or characteristics. In addition, when a specific feature, structure or characteristic is described in conjunction with an embodiment, it should be within the knowledge of a person skilled in the art to implement such feature, structure or characteristic in conjunction with other embodiments (whether or not explicitly described).
[0063] In general, a term can be understood, at least in part, from its use in context. For example, depending, at least in part, on the context, the term "one or more" as used herein can be used to describe any feature, structure, or characteristic in the singular sense, or can be used to describe a combination of features, structures, or characteristics in the plural sense. Additionally, the term "based on" can be understood as not necessarily intended to convey an exclusive set of factors, but can instead, depending, at least in part, on the context, allow for the presence of other factors that are not necessarily explicitly described.
[0064] like Figure 1 As shown, a blueberry yield prediction method comprises the following steps:
[0065] S1: A network of underground sensors installed around the blueberry roots collects real-time data about the roots, including soil moisture, temperature, oxygen content, root conductivity, and soil microbial activity.
[0066] S2: Clean the root data collected in S1, remove abnormal or invalid data, and uniformly process different types of data through standardized methods to ensure that the data have the same dimension, which is convenient for analysis in subsequent steps;
[0067] S3: Based on the data preprocessed by S2, the root health index is calculated using the weighted average method. The root health index is used to comprehensively reflect the effects of multiple factors such as soil moisture, oxygen content, electrical conductivity, and soil microbial activity on root health.
[0068] S4: Fusion of the root health index with environmental data, including meteorological data (soil temperature, precipitation, wind speed, etc.), soil nutrient content, and irrigation water quality. Time alignment algorithms are used to ensure the correlation of different data sources at the same time point, generating a comprehensive feature vector for yield prediction.
[0069] S5: Based on the comprehensive feature vector generated by S4, a blueberry yield prediction model was constructed using the support vector machine algorithm. The support vector machine is suitable for processing multidimensional nonlinear data and is used to predict the yield of blueberries under different root health conditions.
[0070] S6: Use historical blueberry yield data to train the blueberry yield prediction model constructed in S5, and optimize the model parameters through cross-validation to improve the generalization ability and prediction accuracy of the model under different seasons and environmental conditions;
[0071] S7: Based on the real-time collected root health data and environmental data, the trained blueberry yield prediction model is used to make real-time predictions of blueberry yields and generate a yield prediction report.
[0072] S1 specifically includes:
[0073] S11: A sensor network is deployed around the root system in the blueberry planting area. The sensor network consists of a soil moisture sensor, a temperature sensor, an oxygen sensor, a root conductivity sensor, and a soil microbial activity detection sensor. The sensor network is connected to the central data processing unit through a wireless communication module to ensure that the sensor data can be uploaded in real time.
[0074] S12: Use soil moisture sensors to detect changes in moisture content in each layer of the soil and collect soil moisture content data. The sensors are buried in the soil at different depths of the root system to ensure that the data can reflect the moisture conditions of the root system's environment;
[0075] S13: monitor the temperature of the soil around the blueberry root system in real time through a temperature sensor and collect soil temperature data;
[0076] S14: The oxygen content in the soil around the root system is detected by an oxygen sensor, and oxygen content data is collected. The oxygen sensor records the oxygen concentration in real time through a chemical reaction sensor or gas diffusion principle, ensuring that the oxygen supply status of the blueberry root system is accurately reflected;
[0077] S15: The conductivity data of the root system is collected in real time through the root conductivity sensor. The conductivity sensor measures the reaction of the root system to the external current through conductivity, reflecting the changes in root health and nutrient absorption;
[0078] S16: Monitor the activity of microorganisms in the soil through soil microbial activity detection sensors and record microbial activity data; through the above steps, the soil moisture, temperature, oxygen content, root conductivity and soil microbial activity data around the blueberry roots can be comprehensively collected to provide more accurate environmental parameter input for blueberry yield prediction.
[0079] S2 specifically includes:
[0080] S21: Perform a preliminary analysis on the root data collected by S1, and identify abnormal data through threshold detection method and statistical analysis method. The threshold detection method is used to detect the data of each type of sensor according to the preset reasonable data range. For example, for the soil moisture sensor data, the reasonable range of soil moisture content is set to [10%, 50%], and the data exceeding the preset reasonable data range is determined as abnormal data; the statistical analysis method will calculate the mean and standard deviation of each type of data, and eliminate the abnormal data exceeding the range of mean ± 3 times standard deviation; the specific abnormal data detection formula is: D out ={d i ∈D:|d i -μ D |>3σ D}, where d i is the i-th data point, D is the entire data set, μ D is the mean of the data set D, σ D is the standard deviation of data set D;
[0081] S22: Repair abnormal data and missing data. The specific repairs include interpolation method and historical mean method. The interpolation method performs linear interpolation based on the data of adjacent time points. The historical mean method uses the historical mean of the same time period to fill in the missing data. The interpolation method is used to repair continuous data. The formula is:
[0082] Among them, d i (t) is the data repair value at time point t, d i (t-1) and d i (t+1) are the data of adjacent time points t-1 and t+1;
[0083] S23: Standardize different types of data to ensure that different data have the same dimension. The standardization process uses the Z-score standardization method to adjust the mean of each type of data to 0 and the standard deviation to 1 to eliminate the dimensional differences between different data types. The Zscore standardization formula is: Among them, d′ is the standardized data, d i is the original data, μ D is the mean of the data set D, σ D is the standard deviation of data set D;
[0084] S24: After data processing is completed, all data sets are output and compared with historical data to verify the rationality of the data. By comparing the changes in data distribution, it is ensured that the processed data does not contain abnormal points and that the data truly reflects the environment and root status. Through the above steps, the collected root data can be efficiently cleaned, abnormal or invalid data can be eliminated, and through standardization processing, it can be ensured that different types of data have the same dimension, thereby improving the consistency of the data and the accuracy of subsequent yield predictions.
[0085] The calculation of root health index in S3 specifically includes:
[0086] S31: Select characteristic parameters from the data preprocessed by S2, the characteristic parameters include soil moisture, soil temperature, oxygen content, root conductivity and soil microbial activity; and assign weights according to the degree of influence of each characteristic parameter on root health; set the soil moisture weight to w1, the soil temperature weight to w2, the oxygen content weight to w3, the root conductivity weight to w4 and the soil microbial activity weight to w5, and all weights meet the following constraints: w1+w2+w3+w4+w5=1;
[0087] S32: normalize each characteristic parameter to eliminate the order of magnitude differences of different parameters. The normalization adopts the minimum-maximum normalization method to adjust the value of each parameter to the range of [0, 1]. The normalization formula is as follows: Among them, p i′ is the normalized value of the ith characteristic parameter, p i is the original i-th characteristic parameter value, p min is the minimum value of the i-th characteristic parameter, p max is the maximum value of the i-th characteristic parameter;
[0088] S33: Calculate the root health index by weighted average method. The root health index is used to evaluate the overall health of the root system. Specifically, the normalized characteristic parameters are multiplied by the corresponding weights, and the products are added to obtain the root health index H. index , the formula is:
[0089] H index =w1·p1′+w2·p2′+w3·p3′+w4·p4′+w5·p5′, where H index is the calculated root health index, p1′, p2′, p3′, p4′, p5′ are normalized soil moisture, soil temperature, oxygen content, root conductivity and soil microbial activity data respectively, w1, w2, w3, w4, w5 are the weights of the corresponding characteristic parameters; the root health index combines the above different parameters and obtains a total health score through the weighted average method; a higher index indicates that the roots grow in a good environment and are in a healthy state; a lower index may mean that the roots are facing environmental pressure (such as drought, high temperature, lack of oxygen, etc.) or are in poor health and need to be managed and adjusted. Through the above steps, the root health index is calculated using the weighted average method combined with the weights of multiple key parameters, which can comprehensively evaluate the health status of blueberry roots under different environmental conditions and provide accurate and reliable basic data for subsequent yield predictions.
[0090] S4 specifically includes:
[0091] S41: Obtain meteorological data, soil nutrient content data and irrigation water quality data from meteorological stations and sensors. Meteorological data include parameters of temperature, humidity, rainfall and wind speed. Soil nutrient data include the content of nitrogen, phosphorus and potassium. Irrigation water quality data include pH value, salt content and concentration of harmful substances in water. Standardize the environmental data obtained to ensure that they have the same dimension as the root health index. Standardization formula: Among them, e i ′ is the normalized environmental data, e i is the original environmental data, e min and e max They are the minimum and maximum values of the environmental data respectively;
[0092] S42: Based on the timestamps collected from different data sources, the root health index and environmental data are time-aligned. The time alignment algorithm synchronizes data with different collection frequencies to the same time point through interpolation or linear time interpolation to ensure that the root health index and environmental data are consistent within the same time interval. If the data at a certain time point is missing, the interpolation method is used to supplement the data based on the adjacent time points. The time alignment formula is: Among them, e i ′(t) is the alignment data at time point t, e i (t-1) and e i (t+1) is the environmental data at the adjacent time point;
[0093] S43: The time-aligned root health index is integrated with the standardized environmental data to generate a comprehensive feature vector V. The comprehensive feature vector is composed of the root health index, meteorological data, soil nutrient content data and irrigation water quality data, and the expression is: v = [H index ,e1′,e2′,e3′,…,e n ′], where V is the comprehensive feature vector, H index is the root health index, e1′, e2′, e3′,…, e n ′ are normalized environmental data, including meteorological data, soil nutrient data and irrigation water quality data. Through the above steps, the root health index and environmental data can be effectively integrated to ensure that different data sources maintain temporal consistency through the time alignment algorithm, and finally generate a comprehensive feature vector for blueberry yield prediction, thereby improving the accuracy and reliability of model prediction.
[0094] The blueberry yield prediction model constructed by using the support vector machine algorithm in S5 specifically includes:
[0095] S51: Receive the comprehensive feature vector V generated by S4, the feature vector includes the root health index H index and various standardized environmental data e1′, e2′,…, e n ′, and use the feature vector as the input feature data of the model;
[0096] S52: A support vector machine regression algorithm (SVR) is used to construct a blueberry yield prediction model. The support vector machine regression algorithm can handle multidimensional feature vectors and nonlinear feature relationships, and is suitable for situations where blueberry yield is affected by multiple complex factors. The support vector machine regression algorithm maximizes the distance between data points and the boundary by finding a suitable decision boundary, thereby achieving accurate prediction; the support vector machine regression model formula is: f(V)=w·V+b, where f(V) is the blueberry yield prediction value, w is the model's weight vector, b is the bias term, and V is the feature vector; through the above steps, a support vector machine blueberry yield prediction model is constructed to ensure that it can handle multidimensional feature data, providing a basis for subsequent model training and optimization.
[0097] S6 specifically includes:
[0098] S61: Obtain blueberry yield data for many years from historical records as the target variable Y of the training data set, and use the comprehensive feature vector V generated by S5 as input data; each data sample consists of the corresponding historical blueberry yield and the comprehensive feature vector of the corresponding time point; the input data V contains the standardized values of the root health index and environmental data, and the target variable Y is the actual historical blueberry yield; the training data set is specifically represented as: {(V1, Y1), (V2, Y2), ..., (V m , Y m )}, where V i is the feature vector at the i-th time point, Y i is the actual blueberry yield at the i-th time point, and m is the number of samples in the data set;
[0099] S62: Perform model training on the training data set. The goal is to minimize the error between the predicted value f(V) output by the model and the actual output value Y by adjusting the model's weight vector w and bias term b. The model training process also includes using a square error loss function to measure the difference between the predicted value and the actual value. The model training objective function expression is: Among them, w is the model weight vector, b is the bias term, C is the regularization parameter used to control the model's tolerance to errors, ∈ is the loss threshold, and Y i is the actual output value, f(V i ) is the predicted output value;
[0100] S63: In order to improve the generalization ability of the model and avoid overfitting, the cross-validation method is used to optimize the hyperparameters of the model. The hyperparameters include the regularization parameter C and the loss threshold ∈. The cross-validation method specifically divides the training data set into k subsets, selects one subset as the validation set each time, and the remaining k-1 subsets are used to train the model; the cycle is executed k times, the prediction error of each validation set is calculated, and finally the parameter combination that minimizes the error is selected; the formula for cross-validation is: Among them, CV is the cross-validation score, k is the number of cross-validation folds, and m is j is the number of samples for the jth validation, is the actual output of the i-th sample in the j-th validation set, is the predicted value of the i-th sample;
[0101] S64: According to the cross-validation results, the model's hyperparameters C and ∈ are adjusted, and the parameter combination that minimizes the validation error is selected for final model training. The optimized model has stronger generalization ability and can more accurately predict blueberry yield. The expression of the optimized blueberry yield prediction model is: f(V) = w * ·V+b * , where w * is the optimized weight vector, b * is the optimized bias term, V is the comprehensive feature vector, including root health index and environmental data; through the above steps, the historical blueberry yield data is used to train the support vector machine model, and the hyperparameters of the model are optimized by cross-validation method to ensure that the model has good prediction accuracy and generalization ability, and can be effectively applied to the dynamic prediction of blueberry yield.
[0102] S7 specifically includes:
[0103] S71: Generate a real-time feature vector, using the standardized data structure in S2 to combine the real-time collected root health data and environmental data into a real-time comprehensive feature vector V 实时 , the real-time feature vector is expressed as: V 实时 =[H index ,e1′,e2′,e3′,…,e n ′];
[0104] S72: Using the blueberry yield prediction model trained in S6, the real-time comprehensive feature vector V 实时 Input the model to generate real-time blueberry yield forecast value V 实时 , the formula is: 预测 =w * ·V 实时 +b * ;
[0105] S73: Based on the real-time output forecast value V 实时 Generate a forecast report, which includes real-time blueberry yield forecast data, various root health indicators (such as soil moisture, temperature, oxygen content, etc.), real-time environmental data (meteorology, soil nutrients, irrigation water quality) and future trend analysis of blueberry yield. Through the above steps, blueberry yield can be predicted based on real-time data, and a detailed forecast report can be generated to help farm managers make data-driven decisions and ensure efficient management of blueberry growth.
[0106] like Figure 2 As shown, a blueberry yield prediction system is used to implement the above-mentioned blueberry yield prediction method, including the following modules:
[0107] Data acquisition module: used to collect root health data and environmental data in real time through a sensor network installed in the blueberry planting area. Root health data includes soil moisture, soil temperature, oxygen content, root conductivity and soil microbial activity. Environmental data includes meteorological data, soil nutrient content and irrigation water quality. This module sends the collected data to the data preprocessing module through wireless transmission technology;
[0108] Data preprocessing module: used to clean, standardize and normalize the data collected by the data acquisition module. The preprocessed data has a unified dimension and eliminates abnormal and invalid data to ensure that the subsequent models can use consistent data.
[0109] Feature extraction and fusion module: used to extract feature parameters from the pre-processed root health data and environmental data, and generate the root health index through the weighted average method; the feature extraction and fusion module also includes fusing the root health index with the standardized environmental data to generate a comprehensive feature vector as the input of the blueberry yield prediction model;
[0110] Model building and training module: A blueberry yield prediction model is built based on the support vector machine regression algorithm (SVR), and the model is trained and optimized using historical blueberry yield data. The model parameters are adjusted through cross-validation to ensure that the model has high generalization ability and prediction accuracy.
[0111] Real-time prediction module: Based on the real-time comprehensive feature vector generated by the feature extraction and fusion module, the trained blueberry yield prediction model is used to perform real-time prediction of blueberry yield. The real-time prediction results are used as the basis for generating the yield prediction report.
[0112] Report generation module: used to generate a blueberry yield forecast report based on the blueberry yield forecast results output by the real-time forecast module. The forecast report specifically includes real-time predicted yield value, root health status, environmental data and future blueberry yield trend analysis for farm managers to refer to for decision-making.
[0113] A computer device comprises a processor and a memory, wherein a computer program is stored in the memory, and when the computer program is run by the processor, the above-mentioned blueberry yield prediction method is executed.
[0114] The present invention covers any substitution, modification, equivalent method and scheme made on the essence and scope of the present invention. In order to make the public have a thorough understanding of the present invention, specific details are described in detail in the following preferred embodiments of the present invention, but those skilled in the art can fully understand the present invention without the description of these details. In addition, in order to avoid unnecessary confusion about the essence of the present invention, well-known methods, processes, procedures, components and circuits are not described in detail.
[0115] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.
Claims
1. A method for predicting blueberry yield, characterized in that: The following steps are involved: S1: A network of underground sensors installed around the blueberry roots collects real-time root-related data, including soil moisture, temperature, oxygen content, root conductivity, and soil microbial activity; S2: Clean the root data collected by S1, remove abnormal or invalid data, and uniformly process different types of data through standardized methods to ensure that the data have the same dimension; S3: Based on the data preprocessed by S2, the root health index is calculated using the weighted average method. The root health index is used to comprehensively reflect the effects of multiple factors such as soil moisture, oxygen content, electrical conductivity, and soil microbial activity on root health. S4: Fusion of the root health index with environmental data, including meteorological data, soil nutrient content, and irrigation water quality, using a time alignment algorithm to ensure the relevance of different data sources at the same time point, and generating a comprehensive feature vector; S5: Based on the comprehensive feature vector generated by S4, a blueberry yield prediction model was constructed using the support vector machine algorithm to predict the yield of blueberries under different root health conditions; S6: Use historical blueberry yield data to train the blueberry yield prediction model constructed in S5, and optimize the model parameters through cross-validation method; S7: Based on the real-time collected root health data and environmental data, the trained blueberry yield prediction model is used to make real-time predictions on the blueberry yield and generate a yield prediction report.
2. A blueberry yield prediction method according to claim 1, characterized in that: The S1 specifically includes: S11: deploying a sensor network around the root system in the blueberry planting area, wherein the sensor network consists of a soil moisture sensor, a temperature sensor, an oxygen sensor, a root conductivity sensor, and a soil microbial activity detection sensor; S12: Using a soil moisture sensor, detect the moisture content changes in each layer of the soil and collect soil moisture content data; S13: monitor the temperature of the soil around the blueberry root system in real time through a temperature sensor and collect soil temperature data; S14: detecting the oxygen content in the soil around the root system through an oxygen sensor and collecting oxygen content data; S15: collecting the conductivity data of the root system in real time through the root conductivity sensor; S16: Monitor the activity of microorganisms in the soil through soil microbial activity detection sensors and record microbial activity data.
3. A blueberry yield prediction method according to claim 1, characterized in that: The S2 specifically includes: S21: Perform a preliminary analysis on the root system data collected in S1, and identify abnormal data through threshold detection method and statistical analysis method. The threshold detection method is used to detect the data of each type of sensor according to a preset reasonable data range, and determine the data exceeding the preset reasonable data range as abnormal data; the statistical analysis method will calculate the mean and standard deviation of each type of data, and eliminate the abnormal data exceeding the range of mean ± 3 times standard deviation; S22: Repair abnormal data and missing data, including interpolation method and historical mean method; the interpolation method performs linear interpolation based on data of adjacent time points, and the historical mean method uses the historical mean of the same time period to fill in the missing data; S23: Standardize different types of data to ensure that different data have the same dimension. The standardization process uses the Z-score standardization method to adjust the mean of each type of data to 0 and the standard deviation to 1 to eliminate the dimensional differences between different data types. S24: After data processing is completed, all data sets are output and compared with historical data to verify the rationality of the data. By comparing the changes in data distribution, it is ensured that the processed data does not contain abnormal points.
4. A method for predicting blueberry yield according to claim 1, characterized in that: The S3 specifically includes: S31: Select characteristic parameters from the data preprocessed in S2, the characteristic parameters including soil moisture, soil temperature, oxygen content, root conductivity and soil microbial activity; and assign weights according to the degree of influence of each characteristic parameter on root health; set the soil moisture weight to w1, the soil temperature weight to w2, the oxygen content weight to w3, the root conductivity weight to w4 and the soil microbial activity weight to w5, and all weights meet the following constraint: w1+w2+w3+w4+w5=1; S32: normalize each characteristic parameter to eliminate the magnitude differences of different parameters. The normalization adopts the minimum-maximum normalization method to adjust the value of each parameter to the range of [0, 1]. S33: Calculate the root health index by weighted average method. The root health index is used to evaluate the overall health of the root system. Specifically, the normalized characteristic parameters are multiplied by the corresponding weights, and the products are added to obtain the root health index H. index , the formula is: H index =w1·p1′+w2·p2′+w3·p3′+w4·p4′+w5·p5′, where H index is the calculated root health index, and p1′, p2′, p3′, p4′, and p5′ are the normalized soil moisture, soil temperature, oxygen content, root conductivity, and soil microbial activity data, respectively.
5. A method for predicting blueberry yield according to claim 1, characterized in that: The S4 specifically includes: S41: Acquire meteorological data, soil nutrient content data and irrigation water quality data from meteorological stations and sensors, wherein the meteorological data includes parameters of temperature, humidity, rainfall and wind speed, the soil nutrient data includes the content of nutrient elements such as nitrogen, phosphorus and potassium, and the irrigation water quality data includes pH value, salt content and concentration of harmful substances in water; and standardize the acquired environmental data to ensure that they have the same dimension as the root health index; S42: based on the timestamps collected from different data sources, the root health index and the environmental data are time-aligned. The time alignment algorithm synchronizes the data collected at different frequencies to the same time point through interpolation or linear time interpolation to ensure that the root health index and the environmental data are consistent within the same time interval. If the data at a certain time point is missing, the interpolation method is used to supplement the data based on the data at adjacent time points. S43: The time-aligned root health index is fused with the standardized environmental data to generate a comprehensive feature vector V.
6. A method for predicting blueberry yield according to claim 5, characterized in that: The S5 specifically includes: S51: receiving the comprehensive feature vector V generated by S4, and using the feature vector as input feature data of the model; S52: A blueberry yield prediction model is constructed using a support vector machine regression algorithm. The support vector machine regression algorithm achieves accurate prediction by finding a suitable decision boundary and maximizing the distance between data points and the boundary. The support vector machine regression model formula is: f(V)=w·V+b, wherein f(V) is the predicted value of blueberry yield, w is the weight vector of the model, b is the bias term, and V is the feature vector.
7. A method for predicting blueberry yield according to claim 6, characterized in that: The S6 specifically includes: S61: Obtain blueberry production data for many years from historical records as the target variable Y of the training data set, and use the comprehensive feature vector V generated in S5 as input data; the training data set is specifically represented as: {(V1, Y1), (V2, Y2), ..., (V m , Y m )}, where V i is the feature vector at the i-th time point, Y i is the actual blueberry yield at the i-th time point, and m is the number of samples in the data set; S62: Performing model training on the training data set, the goal is to minimize the error between the predicted value f(V) output by the model and the actual output value Y by adjusting the weight vector w and the bias term b of the model; the model training process also includes using a square error loss function to measure the difference between the predicted value and the actual value; S63: In order to improve the generalization ability of the model and avoid overfitting, the cross-validation method is used to optimize the hyperparameters of the model, wherein the hyperparameters include the regularization parameter C and the loss threshold ∈. The cross-validation method specifically divides the training data set into k subsets, selects one subset as the validation set each time, and uses the remaining k-1 subsets for training the model; the loop is executed k times, the prediction error of each validation set is calculated, and finally the parameter combination that minimizes the error is selected; S64: According to the cross-validation results, the model's hyperparameters C and ∈ are adjusted, and the parameter combination that minimizes the validation error is selected for final model training. The optimized blueberry yield prediction model expression is: f(V) = w * ·V+b * , where w * is the optimized weight vector, b * is the optimized bias term, and V is the comprehensive feature vector.
8. A method for predicting blueberry yield according to claim 1, characterized in that: The S7 specifically includes: S71: Generate a real-time feature vector, using the standardized data structure in S2 to combine the real-time collected root health data and environmental data into a real-time comprehensive feature vector V 实时 ; S72: Using the blueberry yield prediction model trained in S6, the real-time comprehensive feature vector V 实时 Input the model to generate real-time blueberry yield forecast value V 实时 ; S73: Based on the real-time output forecast value V 实时 Generate a forecast report that includes real-time blueberry yield forecast data, various root health indicators, real-time environmental data, and future trend analysis of blueberry yield.
9. A blueberry yield prediction system, used to implement a blueberry yield prediction method as claimed in any one of claims 1 to 8, characterized in that: Includes the following modules: Data acquisition module: used to collect root health data and environmental data in real time through a sensor network installed in the blueberry planting area. The root health data includes soil moisture, soil temperature, oxygen content, root conductivity and soil microbial activity. The environmental data includes meteorological data, soil nutrient content and irrigation water quality. Data preprocessing module: used to clean, standardize and normalize the data collected by the data acquisition module; Feature extraction and fusion module: used to extract feature parameters from the pre-processed root health data and environmental data, and generate the root health index through the weighted average method; the feature extraction and fusion module also includes fusing the root health index with the standardized environmental data to generate a comprehensive feature vector; Model building and training module: Build a blueberry yield prediction model based on the support vector machine regression algorithm, and use historical blueberry yield data to train and optimize the model; Real-time prediction module: Based on the real-time comprehensive feature vector generated by the feature extraction and fusion module, the trained blueberry yield prediction model is used to perform real-time prediction of blueberry yield; Report generation module: used to generate a blueberry yield forecast report based on the blueberry yield forecast results output by the real-time forecast module. The forecast report specifically includes real-time predicted yield value, root health status, environmental data and future blueberry yield trend analysis.
10. A computer device, characterized in that: The method comprises a processor and a memory, wherein a computer program is stored in the memory, and when the computer program is executed by the processor, the method for predicting blueberry yield according to any one of claims 1 to 8 is executed.
Citation Information
Patent Citations
Blueberry yield prediction method based on machine learning
CN112906298A
Intelligent greening nursery stock root development tracking system based on big data
CN117035663A
Crop yield prediction algorithm based on multi-source heterogeneous agricultural data
CN117455062A
Greenhouse crop growth monitoring and management system and method based on data analysis
CN118095633A
Tranformer winter wheat yield prediction method in combination with meteorological suitability
CN118485321A
Cited By
Bridge component production multi-protocol fusion intelligent sensing system and modeling method
CN122386988A
