Yield prediction method and device based on mushroom growth state

Abstract

The invention relates to the technical field of image recognition, and discloses a mushroom growth state-based yield prediction method and device, and the method comprises the steps: collecting the time sequence data of mushrooms, determining the two-dimensional feature data and three-dimensional feature data of mushroom sporocarp according to the time sequence data, and carrying out the calculation of the two-dimensional feature data and the three-dimensional feature data; and inputting the two-dimensional feature data and the three-dimensional feature data into a prediction network to calculate a real-time growth rate, when the real-time growth rate is within a normal growth rate range and the mushrooms are in a target growth stage, obtaining a yield prediction result of the mushrooms through a multivariable regression model, and when the real-time growth rate is not within the normal growth rate range, obtaining a yield prediction result of the mushrooms through the multivariable regression model. And determining that the mushrooms are in an abnormal growth state and generating abnormal growth alarm information. Therefore, by implementing the method, the feature recognition precision can be improved through refined image analysis, the reliability and accuracy of growth state judgment and yield prediction are improved, the monitoring accuracy of the mushroom growth state is improved, the yield prediction accuracy and the intelligent degree are further improved, and the labor cost is reduced.

Description

Technical Field

[0001] This invention relates to the field of image recognition technology, and in particular to a method and apparatus for predicting yield based on the growth status of mushrooms. Background Technology

[0002] With the rapid development of facility agriculture, large-scale mushroom cultivation has become an important direction for agricultural industrialization. The market has placed higher demands on the quality stability and yield predictability of mushroom products. In the mushroom cultivation process, real-time monitoring of growth status, advance yield prediction, optimization of planting plans, and adaptation to market demand are core requirements for improving cultivation efficiency and reducing operational risks. However, currently in the mushroom cultivation industry, growth status monitoring largely relies on manual observation or simple equipment testing, and yield prediction lacks accurate data support. This leads to the inability to detect abnormal growth in a timely manner during cultivation, seriously affecting the level of intelligence and economic benefits of cultivation.

[0003] Existing technologies for mushroom yield prediction primarily rely on statistical analysis of historical growth and yield data, using simple trend deductions to arrive at current yield forecasts. This method has significant drawbacks: first, it fails to incorporate real-time growth data, depending solely on historical statistics, leading to low prediction accuracy; second, it cannot dynamically monitor mushroom growth rates, making it difficult to identify abnormal growth in a timely manner, and lacking effective causal analysis and control guidance after abnormalities occur. Therefore, proposing a technical solution that improves the accuracy of mushroom growth status monitoring, thereby enhancing the accuracy and intelligence of yield prediction, is of paramount importance. Summary of the Invention

[0004] This invention provides a method and device for yield prediction based on the growth status of mushrooms, which can help improve the accuracy of mushroom growth status monitoring, thereby improving the accuracy and intelligence of yield prediction.

[0005] To address the aforementioned technical problems, the first aspect of this invention discloses a method for predicting yield based on the growth status of mushrooms, the method comprising: Time-series data of mushrooms are collected at preset time intervals, including spectral image time-series data and point cloud depth data. Two-dimensional feature data of each mushroom fruiting body are determined based on the time-series data of the spectral image, and three-dimensional feature data of each mushroom fruiting body are determined based on the point cloud depth data. The two-dimensional feature data includes the vegetation index, anthocyanin reflectance and fruiting body coordinates of each mushroom fruiting body, and the three-dimensional feature data includes the cap area, stipe length, stipe curvature and space occupancy rate of each mushroom fruiting body. The two-dimensional feature data and the three-dimensional feature data of each mushroom fruiting body are input into the prediction network to calculate the real-time growth rate of the mushroom, and the real-time growth rate is compared with the preset normal growth rate range to obtain the comparison result. When the comparison result indicates that the real-time growth rate is within the normal growth rate range, it is determined whether the mushroom is in the target growth stage. When the mushroom is in the target growth stage, the real-time growth rate and the historical yield data corresponding to the mushroom are input into the multivariate regression model to obtain the yield prediction result of the mushroom. When the comparison result indicates that the real-time growth rate is not within the normal growth rate range, it is determined that the mushroom is in an abnormal growth state, and an abnormal growth alarm message is generated for the abnormal growth state.

[0006] A second aspect of this invention discloses a yield prediction device based on the growth status of mushrooms, the device comprising: The acquisition module is used to acquire time-series data of mushrooms at preset time intervals. The time-series data includes spectral image time-series data and point cloud depth data. The first determining module is used to determine the two-dimensional feature data of each mushroom fruiting body based on the spectral image time series data, and to determine the three-dimensional feature data of each mushroom fruiting body based on the point cloud depth data. The two-dimensional feature data includes the vegetation index, anthocyanin reflectance and fruiting body coordinates of each mushroom fruiting body, and the three-dimensional feature data includes the cap area, stipe length, stipe curvature and space occupancy rate of each mushroom fruiting body. The calculation module is used to input the two-dimensional feature data and the three-dimensional feature data of each mushroom fruiting body into the prediction network, calculate the real-time growth rate of the mushroom, and compare the real-time growth rate with a preset normal growth rate range to obtain the comparison result. The judgment module is used to determine whether the mushroom is in the target growth stage when the comparison result indicates that the real-time growth rate is within the normal growth rate range; The prediction module is used to input the real-time growth rate and the historical yield data of the mushroom into a multivariate regression model when the mushroom is in the target growth stage, so as to obtain the yield prediction result of the mushroom. The first generation module is used to determine that the mushroom is in an abnormal growth state when the comparison result indicates that the real-time growth rate is not within the normal growth rate range, and to generate growth abnormality alarm information for the abnormal growth state.

[0007] As an optional implementation, in a second aspect of the invention, the calculation module inputs the two-dimensional feature data and the three-dimensional feature data of each mushroom fruiting body into the prediction network to calculate the real-time growth rate of the mushroom, specifically including: The two-dimensional feature data and the three-dimensional feature data of each mushroom fruiting body are normalized to obtain a spliced ​​joint feature vector; The predetermined number of joint feature vectors corresponding to each mushroom fruiting body are used to generate a feature matrix in chronological order, and the feature matrix is ​​input into the prediction network to predict the growth increment of each mushroom fruiting body at the next time step. The growth increment includes the increase in cap area and the increase in stipe length. The relative growth rate of area is calculated based on the cap area and the increment of cap area of ​​each mushroom fruiting body; the relative growth rate of length is calculated based on the stipe length and the increment of stipe length of each mushroom fruiting body; and the fruiting body growth rate of each mushroom fruiting body is calculated based on the relative growth rate of area and the relative growth rate of length. The spatial distribution weight of each mushroom fruiting body is determined based on the spatial occupancy rate of each mushroom fruiting body, and the real-time growth rate of the mushroom is calculated based on the fruiting body growth rate and the spatial distribution weight of each mushroom fruiting body.

[0008] As an optional implementation, in the second aspect of the present invention, the prediction module inputs the real-time growth rate and the historical yield data corresponding to the mushroom into a multivariate regression model to obtain the yield prediction result of the mushroom, specifically including: Target historical yield data is selected from the historical yield data corresponding to the mushrooms, and the deviation between the historical growth environment parameters corresponding to the target historical yield data and the current growth environment parameters of the mushrooms is less than a preset deviation threshold. Extract the historical growth rate, historical joint feature vector, and historical yield benchmark value of the mushroom from the target historical yield data, and input the joint feature vector of the mushroom, the real-time growth rate, the historical growth rate, the historical joint feature vector, and the historical yield benchmark value into the multivariate regression model to obtain the predicted value of each mushroom fruiting body. The predicted value of the fruiting body includes the predicted value of the weight of a single fruiting body and the predicted value of the number of fruiting bodies. The historical average prediction error rate of the multivariate regression model is analyzed based on the historical yield data, and the yield prediction result of the mushroom is calculated based on the historical average prediction error rate and the predicted value of each mushroom fruiting body.

[0009] As an optional implementation, in a second aspect of the invention, the apparatus further includes: The first acquisition module is used to acquire sales data of multiple mushroom varieties in the target sales area corresponding to the mushrooms within a historical time period. The sales data includes price fluctuation data of each mushroom variety within the historical time period. The prediction module is also used to predict the market sales forecast data of the mushrooms at the target harvesting stage based on the sales data of each type of mushroom. The market sales forecast data includes the predicted unit price, the transportation unit price, and the packaging material unit price. The first acquisition module is further configured to acquire the cultivation cost data of the mushrooms, and input the market sales forecast data of the mushrooms at the target harvest stage and the cultivation cost data into a preset first profit-maximizing planning model to obtain the predicted optimal profit data for the mushrooms. The predicted optimal profit data includes the optimal harvest time, the optimal harvest quantity, and the predicted profit value.

[0010] As an optional implementation, in a second aspect of the present invention, the first acquisition module is further configured to acquire planting space data of the mushroom cultivation site and the maximum planting area limit for each type of mushroom in the cultivation site, wherein the planting space data includes the maximum effective planting area; The device further includes: The second determining module is used to determine the unit yield prediction profit of each type of mushroom based on the sales data of each type of mushroom, and to determine the unit area yield standard value of each type of mushroom based on the historical yield data. The planning module is used to input the planting space data, the maximum planting area limit for each type of mushroom, the predicted profit per unit yield of each type of mushroom, and the standard value of the yield per unit area for each type of mushroom into a preset second profit-maximizing planning model to obtain the optimal planting data for each type of mushroom, wherein the optimal planting data includes the planting area for each type of mushroom. The prediction module is also used to predict the predicted price trend of each type of mushroom in the target sales area within a future preset time period based on the sales data of various types of mushrooms in the target sales area during a historical time period, and to optimize the optimal planting data of each type of mushroom based on the predicted price trend of each type of mushroom. The second generation module is used to generate a planting plan report based on the optimized optimal planting data and the growth cycle data of each type of mushroom. The planting plan report includes the planting area and batch planting time of each type of mushroom.

[0011] As an optional implementation, in a second aspect of the invention, the apparatus further includes: The retrieval module is used to retrieve the log information of the mushroom's growth environment when the comparison result indicates that the real-time growth rate is not within the normal growth rate range, and to determine whether the abnormal factors causing the mushroom to be in the abnormal growth state include non-environmental abnormal factors based on the log information. The log information includes personnel operation information and inspection information. The second acquisition module is used to acquire the current growth environment parameters of the mushroom when it is determined that the abnormal cause does not include the non-environmental factor abnormal cause. The current growth environment parameters include at least one of temperature parameters, humidity parameters, light parameters, carbon dioxide concentration parameters, and substrate formula parameters. The first determining module is further configured to determine the historical average value of environmental parameters corresponding to the normal growth rate range, and calculate the environmental parameter deviation value based on the current growth environmental parameters and the historical average value of environmental parameters; The first determining module is further configured to match the joint feature vector and the environmental parameter deviation value with historical anomaly records, determine the matching degree score corresponding to each environmental factor anomaly cause, and determine the target environmental factor anomaly cause based on the matching degree score corresponding to each environmental factor anomaly cause. The first generation module is further configured to generate an environmental parameter adjustment amount based on the abnormal growth state and the abnormal inducing factors of the target environmental factors, and to adjust the growth environment of the mushrooms based on the environmental parameter adjustment amount.

[0012] As an optional implementation, in a second aspect of the present invention, the acquisition module is further configured to acquire full data of the cultivation cycle after the cultivation cycle of the mushroom is completed, and store the full data in a preset training database. The full data includes the two-dimensional feature data, the three-dimensional feature data, the growth increment, the real-time growth rate, the yield prediction result, the measured growth increment, the measured growth rate, and the actual harvest yield. The calculation module is also used to calculate the growth increment prediction error of the prediction network and the output prediction error of the multivariate regression model in each piece of the full data when the amount of full data in the training database reaches a preset training trigger number. The device further includes: The training module is used to incrementally train the prediction network and the multivariate regression model using each full data point in the training database and the growth increment prediction error and the yield prediction error corresponding to each full data point, so as to obtain the average growth increment prediction error value corresponding to the prediction network and the average yield prediction error value corresponding to the multivariate regression model. The judgment module is also used to judge whether the average error value of the growth increment prediction and the average error value of the yield prediction meet the preset training accuracy requirements. When the average error value of the growth increment prediction and the average error value of the yield prediction meet the training accuracy requirements, the model optimization information of the trained prediction network and the multivariate regression model is determined. The model optimization information includes parameter configuration information, error optimization magnitude and mushroom species adaptation information. The filtering module is used to store the trained prediction network and the multivariate regression model in the model version management library. When the next batch of mushroom cultivation operation is started, the module filters the target prediction model and target multivariate regression model in the model version management library according to the mushroom variety of the batch of mushrooms, so as to perform the real-time growth rate prediction operation and yield prediction operation of the batch of mushrooms.

[0013] A third aspect of the present invention discloses another yield prediction device based on the growth status of mushrooms, the device comprising: Memory containing executable program code; A processor coupled to the memory; The processor calls the executable program code stored in the memory to execute some or all of the steps in the yield prediction method based on mushroom growth status according to any of the first aspects of the present invention.

[0014] The fourth aspect of the present invention discloses a computer storage medium storing computer instructions, which, when invoked, are used to execute some or all of the steps in the yield prediction method based on the growth status of mushrooms as described in any of the first aspects of the present invention.

[0015] Compared with the prior art, the present invention has the following beneficial effects: In this embodiment of the invention, time-series data of mushrooms are collected at preset time intervals. Two-dimensional feature data of each mushroom fruiting body are determined based on the time-series data of spectral images, and three-dimensional feature data of each mushroom fruiting body are determined based on point cloud depth data. The two-dimensional and three-dimensional feature data of each mushroom fruiting body are input into a prediction network to calculate the real-time growth rate of the mushrooms. The real-time growth rate is compared with a preset normal growth rate range to obtain a comparison result. When the comparison result indicates that the real-time growth rate is within the normal growth rate range, it is determined whether the mushrooms are in the target growth stage. When the mushrooms are in the target growth stage, the real-time growth rate and the corresponding historical yield data of the mushrooms are input into a multivariate regression model to obtain the mushroom yield prediction result. When the comparison result indicates that the real-time growth rate is not within the normal growth rate range, it is determined that the mushrooms are in an abnormal growth state, and an abnormal growth alarm message is generated. It is evident that implementing this invention can improve the accuracy of feature recognition through refined image analysis, achieve full-dimensional coverage of two-dimensional physiological features and three-dimensional morphological features, improve the reliability and accuracy of growth status judgment and yield prediction, realize closed-loop management of growth dynamic monitoring, early yield prediction and timely alarm of anomalies, improve the accuracy of mushroom growth status monitoring, and thus improve the accuracy and intelligence of yield prediction. It can standardize the execution of data collection and processing procedures, reduce labor costs, provide a scientific basis for large-scale breeding decisions, and help improve breeding efficiency. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a flowchart illustrating a method for predicting yield based on the growth status of mushrooms, as disclosed in an embodiment of the present invention. Figure 2 This is a flowchart illustrating another yield prediction method based on the growth status of mushrooms disclosed in an embodiment of the present invention. Figure 3 This is a schematic diagram of the structure of a yield prediction device based on the growth status of mushrooms disclosed in an embodiment of the present invention; Figure 4 This is a schematic diagram of another yield prediction device based on the growth status of mushrooms disclosed in an embodiment of the present invention. Figure 5 This is a schematic diagram of another yield prediction device based on the growth status of mushrooms disclosed in an embodiment of the present invention. Detailed Implementation

[0018] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] The terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this invention are used to distinguish different objects, not to describe a specific order. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, apparatus, product, or end that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or ends.

[0020] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of the invention. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0021] This invention discloses a method and device for yield prediction based on mushroom growth status. It improves feature recognition accuracy through refined image analysis, achieving full-dimensional coverage of two-dimensional physiological features and three-dimensional morphological features. This enhances the reliability and accuracy of growth status judgment and yield prediction. It enables closed-loop management of dynamic growth monitoring, early yield prediction, and timely anomaly alerts, improving the accuracy of mushroom growth status monitoring and thus increasing the accuracy and intelligence of yield prediction. The invention also standardizes data collection and processing procedures, reducing labor costs and providing a scientific basis for large-scale farming decisions, thereby helping to improve farming efficiency. Detailed descriptions follow.

[0022] Example 1 Please see Figure 1 , Figure 1 This is a flowchart illustrating a method for predicting yield based on the growth status of mushrooms, as disclosed in an embodiment of the present invention. Figure 1The described yield prediction method based on mushroom growth status can be applied to a yield prediction device based on mushroom growth status. This yield prediction device can be applied to a mushroom cultivation container. The device can include an intelligent server or intelligent platform for predicting mushroom yield. The intelligent server can include a local server or a cloud server; this embodiment of the invention is not limited thereto. Figure 1 As shown, the yield prediction method based on the growth status of mushrooms may include the following operations: 101. Collect time-series data of mushrooms at preset time intervals.

[0023] In this embodiment of the invention, optionally, the preset time interval represents a fixed time period for collecting mushroom data, which can be adjusted according to different growth stages of the mushrooms. For example, data can be collected every 2 hours during the rapid growth stage and every 6 hours during the mature stage. Growers can adjust this according to different mushroom varieties. The time-series data of the mushrooms represents a dataset with temporal correlation collected continuously at preset time intervals. The time-series data can include spectral image time-series data and point cloud depth data. Spectral image time-series data can be collected using a multispectral camera. The spectral reflectance of different bands can reflect the physiological state of the mushrooms. For example, the near-infrared band can characterize the moisture content of the mushrooms, and the reflectance intensity of a specific band is related to the anthocyanin content. Point cloud depth data of the mushrooms can be collected using lidar or a depth camera. The point cloud depth data can reconstruct the three-dimensional morphology and spatial distribution of the mushrooms. This invention does not limit this.

[0024] 102. Determine the two-dimensional feature data of each mushroom fruiting body based on the time series data of the spectral image, and determine the three-dimensional feature data of each mushroom fruiting body based on the point cloud depth data.

[0025] In this embodiment of the invention, optionally, each mushroom fruiting body may include structures such as a cap and a stipe. The two-dimensional feature data of each mushroom fruiting body may include the vegetation index, anthocyanin reflectance, and fruiting body coordinates. The vegetation index can be calculated using the Normalized Difference Vegetation Index (NDVI) with the formula NDVI = (NIR - R) / (NIR + R), where NIR represents the near-infrared reflectance in the time-series spectral image data, and R represents the red light reflectance. The vegetation index reflects the growth vitality of the mushroom. The anthocyanin reflectance can be obtained by extracting the reflectance intensity of the 550-600nm band in the time-series spectral image data. Anthocyanin content is related to the mushroom's resistance and maturity. Specifically, the first reflectance intensity of the 550-600nm band (the strong absorption band of anthocyanins) in the time-series spectral image data can be extracted. And the second reflection intensity in the 700-800nm ​​band (less affected by anthocyanins) Then calculate the anthocyanin reflectance:

[0026] Indicates the anthocyanin reflectance, first reflectance intensity It is negatively correlated with anthocyanin content, meaning that the lower the anthocyanin reflectance, the higher the anthocyanin content.

[0027] In this embodiment of the invention, optionally, the three-dimensional feature data of each mushroom fruiting body may include the cap area, stipe length, stipe curvature, and space occupancy rate of each mushroom fruiting body. Specifically: cluster analysis can be performed on the point cloud data of each fruiting body to separate the point cloud of the cap portion. The projected outline of the cap is obtained through plane fitting, and the area of ​​this outline is calculated, which is the actual physical area of ​​the cap. In the stipe point cloud data, the connection point between the stipe and the cap (upper end point) and the contact point between the stipe and the substrate (lower end point) can be identified, and the straight-line distance between the two points is calculated, which is the stipe length. Specifically, a point cloud clustering algorithm (cluster radius r = 0.3 cm, minimum number of points N = 50) can be used to segment the point cloud depth data of the mushroom fruiting body to obtain cap point cloud clusters. and stipe dotted clusters The intersection area of ​​two clusters of point clouds center point coordinates This refers to the connection point (upper end point) between the stipe and the cap; the stipe point cloud cluster can be represented by the least squares method. The set of points with the smallest z-coordinate Perform plane fitting, the equation of the fitted plane is: This plane and the cluster of stipe dot clouds intersection This refers to the contact point (lower end point) between the stipe and the substrate; a curve fitting of the stipe's central axis can be used to calculate the mean curvature of the curve. Specifically, this can be done by observing the stipe's point cloud clusters. In the middle, a cross section is extracted every h = 0.5 cm along the z-axis. k is the cross-sectional index, and then the contour extraction algorithm is used to obtain... edge point Then calculate the geometric center coordinates of the edge points. All As feature points, the central axis of the stipe was obtained by fitting a 3rd-order B-spline curve. The equation of the fitted curve is:

[0028] Optionally, a higher curvature indicates a greater degree of stipe bending, while a curvature of 0 indicates that the stipe is completely upright. The ratio of the three-dimensional spatial volume occupied by the point cloud of a single fruiting entity to the total spatial volume of the cultivation area where the fruiting entity is located can be calculated to obtain the space occupancy rate, which reflects the degree of resource occupation of the fruiting entity in the cultivation space. This invention does not impose any limitations on this.

[0029] 103. Input the two-dimensional and three-dimensional feature data of each mushroom fruiting body into the prediction network, calculate the real-time growth rate of the mushroom, and compare the real-time growth rate with the preset normal growth rate range to obtain the comparison results.

[0030] In this embodiment of the invention, optionally, the prediction network can be a time-series data prediction model built based on deep learning, preferably a long short-term memory network or a gated recurrent unit. Its input is the fusion information of two-dimensional feature data and three-dimensional feature data, and its output is the real-time growth rate of the mushroom. Then, the real-time growth rate is compared with a preset normal growth rate range to obtain the comparison result. The preset normal growth rate range can be calculated by a large amount of historical growth data of the same type of mushroom, and the invention does not limit it.

[0031] 104. When the comparison results indicate that the real-time growth rate is within the normal growth rate range, determine whether the mushroom is in the target growth stage. When the mushroom is in the target growth stage, input the real-time growth rate and the corresponding historical yield data of the mushroom into the multivariate regression model to obtain the mushroom yield prediction results.

[0032] In this embodiment of the invention, optionally, the target growth stage refers to the key stage in the mushroom growth process suitable for yield prediction, which is usually the middle of the period after the mycelium matures and the fruiting body expands rapidly. When the mushroom is in the target growth stage, the real-time growth rate and the corresponding historical yield data of the mushroom are input into the multivariate regression model to obtain the yield prediction result of the mushroom.

[0033] 105. When the comparison results indicate that the real-time growth rate is not within the normal growth rate range, it is determined that the mushroom is in an abnormal growth state, and an abnormal growth alarm message is generated.

[0034] In this embodiment of the invention, the abnormal growth status of the mushroom may optionally include growing too slowly or growing too fast. The abnormal growth alarm information may include the time of the abnormality, the current growth rate, the normal range, and the possible impact, etc., and can be pushed to the cultivation management personnel through the display device in the mushroom cultivation container or the cultivation management terminal (such as a computer, mobile APP, etc.).

[0035] It is evident that implementation Figure 1The described yield prediction method based on mushroom growth status can collect time-series data of mushrooms at preset time intervals. It determines the two-dimensional feature data of each mushroom fruiting body based on spectral image time-series data and the three-dimensional feature data of each mushroom fruiting body based on point cloud depth data. The two-dimensional and three-dimensional feature data of each mushroom fruiting body are input into a prediction network to calculate the real-time growth rate of the mushroom. The real-time growth rate is then compared with a preset normal growth rate range to obtain the comparison result. When the comparison result indicates that the real-time growth rate is within the normal growth rate range, it is determined whether the mushroom is in the target growth stage. When the mushroom is in the target growth stage, the real-time growth rate and the corresponding historical yield data are input into a multivariate regression analysis. The model obtains mushroom yield prediction results. When the comparison results indicate that the real-time growth rate is not within the normal growth rate range, it determines that the mushroom is in an abnormal growth state and generates abnormal growth alarm information. It can improve the accuracy of feature recognition through refined image analysis, achieve full-dimensional coverage of two-dimensional physiological features and three-dimensional morphological features, improve the reliability and accuracy of growth status judgment and yield prediction, and realize closed-loop management of growth dynamic monitoring, early yield prediction and timely alarm of anomalies. It improves the accuracy of mushroom growth status monitoring, thereby improving the accuracy and intelligence of yield prediction. It can standardize the execution of data collection and processing processes, reduce labor costs, provide a scientific basis for large-scale breeding decisions, and help improve breeding efficiency.

[0036] In an optional embodiment, the yield prediction method based on mushroom growth status may further include the following operations: Obtain sales data for various mushroom varieties in the target sales region for a historical period, including price fluctuation data for each mushroom variety during the historical period. Based on the sales data of each type of mushroom, predict the market sales data of mushrooms at the target harvest stage. The market sales forecast data includes the predicted unit price, transportation unit price and packaging material unit price. Acquire mushroom cultivation cost data, and input the market sales forecast data and cultivation cost data of mushrooms at the target harvest stage into the preset first profit maximization planning model to obtain the predicted optimal profit data for mushrooms. The predicted optimal profit data includes the optimal harvest time, the optimal harvest quantity, and the predicted profit value.

[0037] In this optional embodiment, the target sales area for the mushrooms can be determined based on information such as the location of the mushroom cultivation site and sales channels. For example, the area covered by local agricultural wholesale markets, supermarkets in surrounding cities, and online e-commerce platforms. The historical time period can include the past 6-12 months to ensure that the data can cover the price fluctuation patterns under different seasons and different supply and demand conditions. The sales data can include the price fluctuation data of each type of mushroom in the historical time period. This embodiment does not limit this.

[0038] In this optional embodiment, the target harvesting stage can optionally represent the time period during which the current batch of mushrooms is expected to mature and be harvested. Market sales forecast data can include the predicted unit price, transportation unit price, and packaging material unit price. Specifically, a time series ARIMA (2,1,2) model can be used to predict the sales unit price during the target harvesting stage. Historical sales data of mushrooms includes time series data collected daily / weekly. The ARIMA (2,1,2) model can eliminate data non-stationarity through differencing and capture price fluctuation trends by combining a second-order autoregressive term and a second-order moving average term. The formula for calculating the predicted unit price of mushrooms during the target harvesting stage includes:

[0039] in, This represents the predicted unit price at time T during the target harvesting stage. The constant term of the model is obtained by fitting historical data, such as... , and Represents the natural regression coefficient, for example , , This represents the first price difference sequence at the first historical moment before the target harvesting stage. This represents the second price difference sequence at the second historical moment before the target harvesting stage, where the price difference sequence is calculated as follows: , Represents the price at time t. This represents the price at time t-1. and Indicates the translation coefficient, for example , , This represents the model prediction error at the first historical moment. This represents the model prediction error at the second historical moment; Since the price trend of mushrooms directly affects the transportation and packaging material demand for mushroom shipments, high mushroom prices indicate high demand, and correspondingly higher transportation and packaging material costs. It is possible to determine the distance between the mushroom cultivation site and the target sales area, as well as the historical transportation cost fluctuation coefficient (e.g., oil price fluctuation coefficient). Based on the historical price fluctuation data for each mushroom variety, the historical price fluctuation coefficient for the current variety can be determined. Then, based on the distance, oil price fluctuation coefficient, and historical price fluctuation coefficient, the unit transportation price of mushrooms can be predicted. The formula for calculating the unit transportation price includes:

[0040] in, This represents the unit price of transporting mushrooms during the target harvesting stage; D represents the distance between the mushroom cultivation site and the target sales area; and F represents the oil price volatility coefficient, with a base value of 1.0. The historical price volatility coefficient of mushrooms is represented by a, b, and c, which are regression coefficients of the model obtained by fitting historical transportation data. For example, a = 0.002, b = 0.3, and c = 0.05. The formula for calculating the historical price volatility coefficient includes:

[0041] in, This represents the highest historical price for the current type of mushroom, recorded from sales data of various mushroom varieties within a historical time period. This indicates the lowest historical price. Indicates the historical average price; The standard usage of packaging materials per unit weight for current mushroom varieties can be determined. Then, based on the standard usage of packaging materials per unit weight and the historical price fluctuation coefficient of mushrooms, the unit price of mushroom packaging materials can be predicted. The formula for calculating the unit price of packaging materials includes:

[0042] in, This represents the unit price of packaging materials for mushrooms, S represents the standard usage of packaging materials per unit weight for the current mushroom category, and d and e represent regression coefficients, for example, d=1.2 and e=0.03; The mushroom cultivation cost data includes the cultivation cost per unit area, specifically including all costs incurred in the process from sowing to harvesting of this batch of mushrooms, including substrate costs (such as the cost of raw materials such as sawdust and cottonseed hulls), energy costs (such as electricity costs for temperature control and lighting equipment), labor costs (such as labor costs for sowing, inspection, and harvesting), equipment depreciation costs (such as the average annual depreciation of equipment such as cultivation chambers and cameras), and pest and disease control costs (such as the cost of fungicides and insecticides), etc. This embodiment does not limit these costs.

[0043] In this optional embodiment, the market sales forecast data and planting cost data of the mushrooms at the target harvest stage can be input into a preset first profit-maximizing planning model to obtain the predicted optimal profit data for the mushrooms. The first profit-maximizing planning model can be a linear programming model, and the predicted optimal profit data includes the optimal harvest time, the optimal harvest quantity, and the predicted profit value. The function of the first profit-maximizing planning model includes:

[0044] The constraints include: , ,

[0045] in, This represents the predicted revenue figure. This indicates the predicted yield of mushrooms, or it can represent the quantity harvested. This represents the cost of planting per unit area, where A represents the planting area. Indicates the maximum production capacity during the breeding cycle. Indicates the maximum available planting space. This represents the optimal harvesting time, which is obtained by solving for the maximum value of the model function. The optimal harvesting time and optimal harvesting quantity can be determined by inputting the predicted unit price, transportation unit price, and packaging material unit price at different times T during the target harvesting stage into the model to obtain the predicted revenue data at different times T, and determining the time corresponding to the maximum predicted revenue data as the optimal harvesting time. This embodiment does not impose any limitations on this.

[0046] As can be seen, implementing this optional embodiment can obtain sales data of various mushroom varieties in the target sales area over a historical period. Based on the sales data of each mushroom variety, it can predict market sales forecast data for the target harvest stage, obtain mushroom cultivation cost data, and input the market sales forecast data and cultivation cost data into a preset first profit-maximizing planning model to obtain the predicted optimal profit data for mushrooms. This can achieve synergistic optimization of yield and profit, improve the economic benefits of aquaculture, accurately capture dynamic market price trends, avoid decision-making biases caused by fixed price estimations, improve the reliability and adaptability of profit forecasts, and transform abstract profit analysis into directly applicable harvesting decision guidelines. This solves the problem of the disconnect between profit analysis and actual aquaculture operations, covers the core decision-making needs of the entire aquaculture cycle, further enhances the intelligence and marketization level of mushroom aquaculture, and helps farmers maximize their profits.

[0047] In another optional embodiment, the yield prediction method based on the growth status of mushrooms may further include the following operations: Obtain planting space data of mushroom cultivation sites and the maximum planting area limit for each type of mushroom in the cultivation site. The planting space data includes the maximum effective planting area. The projected profit per unit yield for each type of mushroom is determined based on sales data for each type of mushroom, and the standard yield per unit area for each type of mushroom is determined based on historical yield data. The planting space data, the maximum planting area limit for each type of mushroom, the predicted profit per unit yield for each type of mushroom, and the standard value of the yield per unit area for each type of mushroom are input into the preset second profit-maximizing planning model to obtain the optimal planting data for each type of mushroom. The optimal planting data includes the planting area for each type of mushroom. Based on the sales data of various mushroom varieties in the target sales area over a historical period, predict the price trend of each mushroom variety in the target sales area over a future preset period, and optimize the optimal planting data for each mushroom variety based on the predicted price trend of each mushroom variety. Based on the optimized planting data and the growth cycle data of each mushroom variety, a planting plan report is generated, which includes the planting area and batch planting time for each mushroom variety.

[0048] In this optional embodiment, the planting space data of the mushroom cultivation site may include the maximum effective planting area. The maximum planting area limit for each type of mushroom represents the upper limit of the planting area set for each type of mushroom based on factors such as the equipment adaptability of the cultivation site, the compatibility of mushroom growth, and the saturation of market demand. The maximum planting area limit for each type of mushroom is less than or equal to the maximum effective planting area of ​​the mushroom cultivation site. This embodiment does not impose any limitation on this.

[0049] In this optional embodiment, the optimal planting data for each type of mushroom represents the planning data for subsequent mushroom planting after the current batch of mushroom planting is completed. The optimal planting data includes the planting area for each type of mushroom and represents the planting data when the sum of the expected returns for each type of mushroom is maximized. That is, the optimal planting data can maximize the returns. The preset function of the second return-maximizing planning model includes:

[0050] The constraints include: ,

[0051] in, This represents the expected total revenue output by the second-maximum-return planning model, where j represents the index of the mushroom category. This represents the planting area of ​​the j-th type of mushroom. This represents the standard yield per unit area for the j-th type of mushroom. This represents the predicted profit per unit yield of the j-th type of mushroom. This represents the maximum planting area limit for the j-th type of mushroom. This represents the maximum effective planting area, i.e., the maximum available planting space. The optimal planting data for each mushroom variety includes the planting area for each mushroom variety that maximizes the expected total revenue output by the second revenue-maximizing planning model. Specifically, multiple combinations of planting areas for different mushroom varieties can be determined, for example, combination A: the planting area for shiitake mushrooms is... 30% of the area is used for oyster mushroom cultivation. 40% of the area is used for enoki mushroom cultivation. 30%; Combination B: Shiitake mushroom cultivation area is 30% of the area is used for oyster mushroom cultivation. 20% of the area is used for enoki mushroom cultivation. 50%; to obtain N combinations, then input the mushroom planting area of ​​each combination into the model to obtain N expected total revenues, among which the combination with the largest total revenue is the mushroom planting area.

[0052] In this optional embodiment, the predicted price trend of each type of mushroom within a preset future time period is matched with the mushroom's growth cycle, preferably 30-60 days. A triple exponential smoothing method can be used to predict the predicted price trend of each type of mushroom within the target sales area within the preset future time period. Specifically, the horizontal smoothing of historical sales data for the j-th type of mushroom can be calculated. Trend smoothing and seasonal smoothness ,in:

[0053]

[0054]

[0055] The formula for predicting the price of type j mushrooms over the next h time points includes:

[0056] in, , , These represent the horizontal smoothing coefficient, trend smoothing coefficient, and seasonal smoothing coefficient, respectively. s represents the seasonal cycle length of the j-th type of mushroom, and h represents the preset step size, i.e., the number of future time nodes that need to be predicted. The optimal planting data for each mushroom type can be optimized based on the predicted price trend. Then, a planting plan report is generated based on the growth cycle data of each mushroom type. The planting plan report guides growers in performing termination operations. The planting plan report can include the planting area and batch planting time for each mushroom type. Specifically, it can include the planting area, sowing time, transplanting time, expected harvest time, expected yield, and expected profit for each mushroom type. This embodiment does not limit this.

[0057] As can be seen, implementing this optional embodiment can obtain planting space data of mushroom cultivation sites and the maximum planting area limit for each type of mushroom. Based on sales data for each type of mushroom, the predicted profit per unit yield is determined, and based on historical yield data, the standard value of yield per unit area is determined. The planting space data, maximum planting area limit, predicted profit per unit yield, and standard value of yield per unit area are input into a second profit-maximizing planning model to obtain the optimal planting data for each type of mushroom. The optimal planting data is further optimized by combining the predicted price trend of each type of mushroom in the target sales area within a preset future time period. A planting plan report is generated based on the optimized planting data and the growth cycle data of each type of mushroom. This enables the scientific optimization of planting space and mushroom types, avoiding resource waste and the risk of over-cultivation of a single type, improving space utilization, dynamically adapting to market price fluctuations, enhancing the risk resistance and market adaptability of the planting plan, avoiding price drops caused by concentrated market entry through staggered planting and harvesting times, ensuring stable breeding income, generating standardized and structured planting plan reports, reducing the difficulty of breeding decisions, and further improving the stability of large-scale breeding benefits and the level of intelligent management by rationally combining multiple types of mushrooms to diversify market risks.

[0058] In yet another optional embodiment, the yield prediction method based on the growth status of mushrooms may further include the following operations: After completing the cultivation cycle for the mushrooms, all data of the cultivation cycle are collected and stored in a preset training database. The full data includes the two-dimensional feature data, the three-dimensional feature data, the growth increment, the real-time growth rate, the yield prediction result, the measured growth increment, the measured growth rate, and the actual harvest yield. When the total amount of data in the training database reaches the preset number of training triggers, calculate the growth increment prediction error of the prediction network and the output prediction error of the multivariate regression model in each set of data. By training each full dataset in the training database and the corresponding growth increment prediction error and yield prediction error, the prediction network and multivariate regression model are incrementally trained to obtain the average growth increment prediction error value of the prediction network and the average yield prediction error value of the multivariate regression model. Determine whether the average error value of the growth increment prediction and the average error value of the yield prediction meet the preset training accuracy requirements. When the average error value of the growth increment prediction and the average error value of the yield prediction meet the training accuracy requirements, determine the model optimization information of the trained prediction network and the multivariate regression model. The model optimization information includes parameter configuration information, error optimization magnitude and mushroom species adaptation information. The trained prediction network and multivariate regression model are stored in the model version management library. When the next batch of mushroom cultivation is started, the target prediction model and target multivariate regression model are selected from the model version management library according to the mushroom variety of the batch, so as to perform real-time growth rate prediction and yield prediction operations for the batch of mushrooms.

[0059] In this optional embodiment, the cultivation cycle can refer to the complete time period from mushroom inoculation to mature harvesting. The cultivation cycle varies for different mushroom varieties. The full data of the cultivation cycle can include two-dimensional feature data and three-dimensional feature data of the mushroom, the growth increment, the real-time growth rate, the yield prediction result, the measured growth increment, the measured growth rate, and the actual harvest yield. It can also include the growth environment parameters of the mushroom. The actual harvest yield can include the actual total yield, the average weight of a single fruiting body, the actual number of fruiting bodies, etc. This embodiment does not limit this.

[0060] In this optional embodiment, the training database is optionally used to store the full data corresponding to each breeding cycle. When the number of full data in the training database reaches a preset training trigger number, such as 50, the growth increment prediction error of the prediction network and the yield prediction error of the multivariate regression model in each full data can be calculated. Specifically, the growth increment prediction error represents the absolute value of the difference between the growth increment and the measured growth increment. The growth increment prediction error may include the cap area increment prediction error and the stipe length increment prediction error. The yield prediction error represents the absolute value of the difference between the yield prediction result output by the multivariate regression model and the actual harvest yield in that cycle. This embodiment does not limit this.

[0061] In this optional embodiment, the prediction network and multivariate regression model can be incrementally trained using each full dataset in the training database and the corresponding growth rate prediction error and yield prediction error. The average incremental growth prediction error value of the prediction network represents the average growth rate prediction error of the prediction network on the new data after incremental training, and the average yield prediction error value of the multivariate regression model represents the average yield prediction error of the multivariate regression model on the new data after incremental training. It can be determined whether the average incremental growth prediction error value and the average yield prediction error value meet the preset training accuracy requirements. If they do, the model optimization information of the trained prediction network and multivariate regression model can be determined. The model optimization information may include parameter configuration information, error optimization magnitude, and mushroom species adaptation information. The mushroom species adaptation information indicates the mushroom species to which the optimized model is applicable (e.g., only applicable to oyster mushrooms, or applicable to both oyster mushrooms and enoki mushrooms). Each optimized model version is different, and different versions of the model have different prediction effects on different types of mushrooms. This avoids the decrease in prediction accuracy caused by cross-species use. This embodiment does not impose any limitations.

[0062] In this optional embodiment, the model version management library can be used to store databases of different versions of optimized models, categorized by mushroom type and optimization time, recording the model file, model optimization information, and applicable scenarios for each version, facilitating subsequent calls and version backtracking. When the next batch of mushroom cultivation operations is started, the target prediction model and target multivariate regression model can be selected from the model version management library according to the mushroom type of the batch of mushrooms to perform real-time growth rate prediction and yield prediction operations for the batch of mushrooms. This embodiment does not impose any limitations.

[0063] As can be seen, implementing this optional embodiment can collect all data for the mushroom cultivation cycle after completion and store it in a preset training database. When the amount of all data in the training database reaches the preset training trigger number, the growth rate prediction error of the prediction network and the yield prediction error of the multivariate regression model in each piece of all data are calculated. The two models are incrementally trained using the all data in the training database and the corresponding prediction errors. When the training accuracy requirements are met, the model optimization information is determined and stored in the model version management library. In the next batch of cultivation, the target model matching the mushroom category is selected and the prediction operation is performed. It can continuously supplement new data through incremental training, solve the problem of the decline in prediction accuracy after long-term use of traditional models, and achieve continuous improvement in model accuracy. It can specifically correct model biases, improve optimization efficiency and the extent of improvement in prediction accuracy. It can achieve accurate adaptation of the model to different mushroom categories and cultivation scenarios, avoid prediction distortion caused by cross-scenario use, improve the reusability and implementation efficiency of the technical solution, reduce the labor cost of model maintenance, adapt to new situations such as changes in the cultivation environment and mushroom variety improvement, and ensure the long-term stable accuracy of the entire prediction method, providing reliable technical support for the continuous intelligent upgrading of mushroom cultivation.

[0064] Example 2 Please see Figure 2 , Figure 2 This is a flowchart illustrating a method for predicting yield based on the growth status of mushrooms, as disclosed in an embodiment of the present invention. Figure 2 The described yield prediction method based on mushroom growth status can be applied to a yield prediction device based on mushroom growth status. This yield prediction device can be applied to a mushroom cultivation container. The device can include an intelligent server or intelligent platform for predicting mushroom yield. The intelligent server can include a local server or a cloud server; this embodiment of the invention is not limited thereto. Figure 2 As shown, the yield prediction method based on the growth status of mushrooms may include the following operations: 201. Collect time-series data of mushrooms at preset time intervals.

[0065] 202. Determine the two-dimensional feature data of each mushroom fruiting body based on the time series data of the spectral image, and determine the three-dimensional feature data of each mushroom fruiting body based on the point cloud depth data.

[0066] 203. Normalize the two-dimensional and three-dimensional feature data of each mushroom fruiting body to obtain the spliced ​​joint feature vector.

[0067] In this embodiment of the invention, optionally, two-dimensional feature data and three-dimensional feature data of different dimensions can be converted to the same numerical range. Specifically, the Min-Max normalization algorithm can be used to obtain the spliced ​​joint feature vector. This vector integrates the planar morphology, physiological characteristics, three-dimensional morphology and spatial distribution information of the sub-entity, providing a foundation for subsequent temporal feature construction.

[0068] 204. Generate a feature matrix by taking the preset number of joint feature vectors corresponding to each mushroom fruiting body in chronological order, and input the feature matrix into the prediction network to predict the growth increment of each mushroom fruiting body at the next time step.

[0069] In this embodiment of the invention, optionally, the preset number can be determined based on the preset time interval of data collection and the sensitivity of growth rate change, preferably 5-10. For example, if the collection interval is 6 hours and the preset number is 8, then it covers the temporal characteristics of 48 hours. The growth increment can include the increase in cap area and the increase in stipe length, which is not limited in this invention.

[0070] 205. Calculate the relative growth rate of area based on the cap area and cap area increment of each mushroom fruiting body, calculate the relative growth rate of length based on the stipe length and stipe length increment of each mushroom fruiting body, and calculate the fruiting body growth rate of each mushroom fruiting body based on the relative growth rate of area and the relative growth rate of length.

[0071] In this embodiment of the invention, optionally, the relative growth rate of area can represent the ratio of the increase in cap area to the current cap area, reflecting the relative growth rate of the cap and avoiding misjudgment of growth rate due to differences in initial cap area. For example, small caps may have a smaller absolute increase but a higher relative growth rate. The relative growth rate of length represents the ratio of the increase in stipe length to the current stipe length. Combining the relative growth rate of area and the relative growth rate of length, the growth rate of a single fruiting body can be calculated. Specifically, a weighted average calculation can be used, with the weights determined according to the mushroom variety. The specific formula for calculating the fruiting body growth rate is as follows:

[0072] in, This represents the fruiting body growth rate of the i-th mushroom fruiting body. This represents the relative growth rate of the area of ​​the i-th mushroom fruiting body. This represents the relative growth rate of the length of the i-th mushroom fruiting body. The weighting coefficients represent the relative growth rate of area. The weighting coefficients represent the relative growth rate of length. The value can be determined based on historical data for different types of mushrooms. For example, the growth of the cap of shiitake mushrooms has a greater impact on yield, so the weighting coefficient for the relative growth rate of area is 0.6 and the weighting coefficient for the relative growth rate of length is 0.4. This invention does not impose any limitations on this.

[0073] 206. Determine the spatial distribution weight of each mushroom fruiting body based on its spatial occupancy rate, and calculate the real-time growth rate of the mushroom based on its fruiting body growth rate and spatial distribution weight.

[0074] In this embodiment of the invention, optionally, the spatial distribution weight of each mushroom fruiting body can be determined based on its spatial occupancy rate. This weight value characterizes the degree of influence of the fruiting body on the overall growth rate. A higher spatial occupancy rate indicates that the fruiting body occupies more cultivation resources (space, nutrients, light), and has a greater impact on the overall growth state, resulting in a higher weight value. Then, based on the fruiting body growth rate and spatial distribution weight of each mushroom fruiting body, the real-time growth rate of the entire mushroom is calculated. The formula for calculating the real-time growth rate is as follows:

[0075] in, This represents the real-time growth rate of the mushrooms, where n represents the total number of mushroom fruiting bodies in the mushroom cultivation area. This represents the spatial distribution weight of the i-th mushroom fruiting entity, and the sum of the spatial distribution weights of all mushroom fruiting entities is 1.

[0076] 207. Compare the real-time growth rate with the preset normal growth rate range to obtain the comparison results.

[0077] 208. When the comparison results indicate that the real-time growth rate is within the normal growth rate range, determine whether the mushroom is in the target growth stage. When the mushroom is in the target growth stage, input the real-time growth rate and the corresponding historical yield data of the mushroom into the multivariate regression model to obtain the mushroom yield prediction results.

[0078] 209. When the comparison results indicate that the real-time growth rate is not within the normal growth rate range, it is determined that the mushroom is in an abnormal growth state, and an abnormal growth alarm message is generated.

[0079] In this embodiment of the invention, for other descriptions of steps 201, 202 and steps 207-209, please refer to the detailed description of steps 101-105 in Embodiment 1 of the invention. These descriptions will not be repeated in this embodiment of the invention.

[0080] It is evident that implementation Figure 2The described yield prediction method based on mushroom growth status can collect time-series data of mushrooms at preset time intervals. It determines the two-dimensional feature data of each mushroom fruiting body based on spectral image time-series data and the three-dimensional feature data of each fruiting body based on point cloud depth data. The two-dimensional and three-dimensional feature data of each mushroom fruiting body are normalized to obtain a spliced ​​joint feature vector. A preset number of joint feature vectors corresponding to each mushroom fruiting body are used to generate a feature matrix in chronological order and input into a prediction network to predict growth increments. The relative growth rates of area and length are calculated based on cap area, stipe length, and corresponding increments to obtain the fruiting body growth rate. The real-time growth rate of the mushroom is calculated by combining spatial distribution weights determined by spatial occupancy rate. This method can eliminate interference from features of different dimensions, ensure the data consistency and reliability of the joint feature vectors, accurately capture the dynamic changes in mushroom growth, take into account the growth characteristics of different parts of the mushroom and the differences in spatial resource occupancy of fruiting bodies, and avoid misjudgments of growth rate caused by single indicators or unweighted calculations. It can improve the accuracy and scientific nature of real-time growth rate calculations through standardized and refined calculation processes. The method also considers the growth characteristics of different parts of the mushroom and the differences in spatial resource occupancy of fruiting bodies, avoiding misjudgments of growth rate caused by single indicators or unweighted calculations. The system inputs 2D and 3D feature data into a prediction network to calculate the real-time growth rate of mushrooms. This real-time growth rate is then compared with a preset normal growth rate range. If the comparison result indicates that the real-time growth rate is within the normal range, it determines whether the mushroom is in the target growth stage. If the mushroom is in the target growth stage, the real-time growth rate and corresponding historical yield data are input into a multivariate regression model to obtain the mushroom yield prediction result. If the comparison result indicates that the real-time growth rate is outside the normal range, it determines that the mushroom is in an abnormal growth state and generates an abnormal growth alarm message. This system can improve feature recognition accuracy through refined image analysis, achieving full-dimensional coverage of 2D physiological features and 3D morphological features, thus improving the reliability and accuracy of growth state judgment and yield prediction. It enables closed-loop management of dynamic growth monitoring, early yield prediction, and timely anomaly alarms, improving the accuracy of mushroom growth state monitoring, thereby increasing the accuracy and intelligence of yield prediction. It also standardizes data collection and processing procedures, reduces labor costs, provides a scientific basis for large-scale farming decisions, and helps improve farming efficiency.

[0081] In an optional embodiment, inputting real-time growth rate and corresponding historical yield data of mushrooms into a multivariate regression model to obtain mushroom yield prediction results may include the following operations: Filter the target historical yield data from the historical yield data of mushrooms. The deviation between the historical growth environment parameters corresponding to the target historical yield data and the current growth environment parameters of mushrooms is less than the preset deviation threshold. Extract the historical growth rate, historical joint feature vector, and historical yield benchmark value of mushrooms from the target historical yield data. Input the joint feature vector, real-time growth rate, historical growth rate, historical joint feature vector, and historical yield benchmark value of mushrooms into a multivariate regression model to obtain the predicted value of each mushroom fruiting body. The predicted value of the fruiting body includes the predicted value of the weight of a single fruiting body and the predicted value of the number of fruiting bodies. Based on historical yield data analysis, the average historical prediction error rate of the multivariate regression model is analyzed, and the yield prediction results of mushrooms are calculated based on the average historical prediction error rate and the predicted value of each mushroom fruiting body.

[0082] In this optional embodiment, the deviation between the historical growth environment parameters corresponding to the target historical yield data and the current growth environment parameters of the mushroom is less than a preset deviation threshold. The growth environment parameters may include one or more combinations of temperature parameters, humidity parameters, light intensity parameters, carbon dioxide concentration parameters, substrate formula parameters, etc. The predicted value of each mushroom fruiting body may include a predicted value of the weight of a single fruiting body and a predicted value of the number of fruiting bodies. The predicted value of the weight of a single fruiting body represents the expected weight of each fruiting body after maturity, and the predicted value of the number of fruiting bodies represents the total number of fruiting bodies that can be harvested in the current cultivation area. This embodiment does not limit this.

[0083] In this optional embodiment, the historical average prediction error rate of the multivariate regression model can be analyzed based on historical production data. The historical average prediction error rate refers to the average error between the predicted production and the actual production in the historical application of the multivariate regression model, reflecting the historical prediction accuracy of the model. The preliminary production prediction result of mushrooms can be obtained by multiplying the predicted weight of a single fruiting body and the predicted number of fruiting bodies. Then, the preliminary production prediction result is corrected by the historical average prediction error rate of the multivariate regression model to obtain the final production prediction result of mushrooms. Specifically: The formula for calculating the preliminary yield forecast of mushrooms is as follows:

[0084] in, This indicates the preliminary production forecast (in kg). This represents the predicted weight of a single fruiting body of the i-th mushroom fruiting body. This represents the predicted number of fruiting bodies of the i-th mushroom fruiting body; Formulas for correcting preliminary yield forecasts using the historical average error rate of multivariate regression models include:

[0085] in, This indicates the predicted yield of mushrooms. This represents the historical average prediction error rate of a multivariate regression model, ranging from 0 to 1. It is taken as a negative number when the model's average predicted value is higher than the actual average value for only a few periods, and as a positive number when it is lower than the actual average value. Specifically, the formula for calculating the historical average prediction error rate includes:

[0086] Where m represents the number of historical mushroom cultivation cycles used to analyze the average error rate of historical predictions in the multivariate regression model, and each cycle includes the predicted yield output by the corresponding multivariate regression model and the actual yield recorded. This represents the predicted output for the k-th period. This represents the actual output in the k-th period. For example, when m=3, , , , , , If the average predicted value of 105 kg is less than the average actual value of 110 kg, then the average error rate of historical predictions is taken as a positive number. This embodiment does not impose any limitation on this.

[0087] As can be seen, implementing this optional embodiment can filter out target historical yield data from the historical yield data corresponding to mushrooms, where the deviation between historical growth environment parameters and current growth environment parameters is less than a preset threshold. Historical growth rate, historical joint feature vector, and historical yield benchmark value are extracted from the target historical yield data. The mushroom's joint feature vector, real-time growth rate, historical growth rate, historical joint feature vector, and historical yield benchmark value are input into a multivariate regression model to obtain the predicted value of the offspring entity. The yield prediction result of the mushroom is calculated by combining the historical prediction average error rate. This ensures a high degree of adaptability between historical data and the current cultivation environment, avoids interference from irrelevant historical data, integrates multi-dimensional core features as model input, improves the scientificity and comprehensiveness of yield prediction, corrects preliminary prediction results through the historical prediction average error rate, effectively offsets inherent model biases, significantly improves the accuracy and reliability of yield prediction, and outputs detailed predicted values ​​for the weight and quantity of individual offspring entities, expanding the application scenarios of the prediction results and providing data support for refined management such as harvesting arrangements and graded sales, further enhancing the scientific nature of mushroom cultivation decisions.

[0088] In another optional embodiment, when the comparison results indicate that the real-time growth rate is not within the normal growth rate range, the yield prediction method based on the mushroom growth status may further include the following operations: Retrieve log information of the mushroom's growth environment and determine whether the abnormal factors causing the mushroom's abnormal growth state include non-environmental abnormal factors based on the log information. The log information includes personnel operation information and inspection information. When it is determined that the abnormal causes do not include non-environmental factors, the current growth environment parameters of the mushrooms are obtained. The current growth environment parameters include at least one of the following: temperature parameters, humidity parameters, light parameters, carbon dioxide concentration parameters, and substrate formula parameters. Determine the historical average values ​​of environmental parameters corresponding to the normal growth rate range, and calculate the environmental parameter deviation values ​​based on the current growth environmental parameters and the historical average values ​​of environmental parameters; The joint feature vector and environmental parameter deviation values ​​are matched with historical anomaly records to determine the matching degree score corresponding to each environmental factor anomaly cause. Based on the matching degree score corresponding to each environmental factor anomaly cause, the target environmental factor anomaly cause is determined. Based on abnormal growth status and abnormal inducing factors of target environmental factors, environmental parameter adjustment quantities are generated, and the growth environment of mushrooms is adjusted according to the environmental parameter adjustment quantities.

[0089] In this optional embodiment, the log information of the mushroom growth environment may include personnel operation information and inspection information. Personnel operation information may include the operation records of the planting managers on the equipment, such as whether the set temperature of the temperature control equipment has been adjusted, whether the substrate moisture has been added, and whether the ventilation equipment has been turned on. Inspection information may include the regular inspection content of the planting managers, such as whether pests and diseases have been found in the mushrooms, whether the cultivation equipment is operating normally, and whether the substrate is moldy. Non-environmental abnormal factors refer to factors that cause abnormal growth of mushrooms and are unrelated to environmental parameters, including cultivation equipment failure (such as temperature control equipment failure leading to temperature out-of-control, and lighting equipment damage leading to insufficient light), human operation errors (such as incorrect adjustment of environmental parameters, incorrect substrate ratio), and pest and disease attacks (such as mold infection, and pests eating fruiting bodies). These types of factors cannot be solved by adjusting environmental parameters and need to be investigated separately. Specifically, the determination of human operation errors can be achieved through log comparison and factor elimination. First, retrieve the personnel operation information during the abnormal period and compare it with the preset standard operation procedure. For example, compare the temperature value of the temperature adjustment operation with the standard temperature adjustment range, and then determine the operation deviation. When the operation deviation is greater than the preset operation deviation threshold, for example, the temperature operation deviation threshold can be 2 degrees Celsius, and the time difference between the personnel operation time and the abnormal growth start time is less than 24 hours, and the inspection information confirms that there is no equipment failure, such as a temperature adjustment equipment failure, it can be determined that the operation was wrong. This embodiment does not limit this.

[0090] In this optional embodiment, when it is determined that the abnormal cause does not include non-environmental abnormal causes, the current growth environment parameters of the mushroom can be obtained. The current growth environment parameters may include at least one of temperature parameters, humidity parameters, light parameters, carbon dioxide concentration parameters, and substrate formula parameters. The historical average value of the environmental parameters corresponding to the normal growth rate range can represent the average value of each environmental parameter when the mushroom growth rate is within the normal range, based on historical data statistics. Then, the environmental parameter deviation value is calculated based on the current growth environment parameters and the historical average value of the environmental parameters. This embodiment does not limit this.

[0091] In this optional embodiment, the joint feature vector and environmental parameter deviation values ​​are optionally matched with historical anomaly records. The historical anomaly records may include multiple historical anomaly cases, and each historical anomaly case includes a corresponding historical joint feature vector, historical environmental parameter deviation values, and at least one corresponding environmental factor anomaly trigger. Specific matching steps include: The joint feature vector and the environmental parameter deviation value can be normalized to obtain the standardized joint feature vector and the standardized environmental parameter deviation value. The historical joint feature vector and historical environmental parameter deviation value of each historical anomaly case in the historical anomaly record are normalized to obtain the standardized historical joint feature vector and standardized historical environmental parameter deviation value. Calculate the matching degree between the standardized joint feature vector and the standardized environmental parameter deviation value and each historical anomaly case, and determine the historical anomaly cases with a matching degree greater than the preset matching degree threshold as target historical anomaly cases; Based on each environmental factor anomaly trigger corresponding to each target historical anomaly case, the frequency of occurrence of each environmental factor anomaly trigger is counted. Based on the frequency of occurrence of each environmental factor anomaly trigger and the matching degree of the target historical anomaly case to which the environmental factor anomaly trigger belongs, the matching degree score corresponding to each environmental factor anomaly trigger is calculated.

[0092] Optionally, the joint feature vector, environmental parameter deviation value, historical joint feature vector of each historical anomaly case in the historical anomaly record, and historical environmental parameter deviation value can be normalized to the [0,1] interval to eliminate the matching distortion caused by the different dimensions between parameters. Optionally, the standardized joint feature vector and the standardized environmental parameter deviation values ​​can be integrated into a fused feature vector according to a preset fusion weight. The formula for the fused feature vector includes:

[0093] The standardized historical joint feature vector and standardized historical environmental parameter deviation value of each historical anomaly case can be integrated into a historical fusion feature vector according to preset fusion weights. The formula for the historical fusion feature vector includes:

[0094] in, Represents the joint feature weights. Indicates the weight of environmental parameter deviations. Represents the fused feature vector. Represents the standardized joint feature vector. Indicates the deviation value of standardized environmental parameters. This represents the historical fusion feature vector of the k-th historical anomaly case. This represents the standardized historical joint feature vector of the k-th historical anomaly case. This represents the standardized historical environmental parameter deviation value for the k-th historical anomaly case; Optionally, the matching degree between the fused feature vector and the historical fused feature vector of each historical anomaly case can be calculated. The matching degree calculation formula includes:

[0095] in, This indicates the degree of matching with the k-th historical anomaly case; Optionally, historical anomaly cases with a matching degree greater than a preset matching degree threshold can be selected as target historical anomaly cases. Then, the frequency of occurrence of each environmental factor anomaly trigger can be counted, and the formula for calculating the matching degree score corresponding to each environmental factor anomaly trigger includes:

[0096] in, This represents the matching score of the abnormal environmental factor identified as 's'. This indicates the number of target historical anomaly cases that contain environmental factor anomaly triggers identified as 's'. This represents the matching degree of the h-th target historical anomaly case among target historical anomaly cases containing environmental factor anomaly triggers identified as s. This indicates the frequency of occurrence of abnormal environmental factors identified as 's'.

[0097] Optionally, a higher matching score indicates that the historical cause is more likely to be the cause of the current anomaly. Then, based on the matching score corresponding to each environmental factor anomaly cause, the target environmental factor anomaly cause is determined. Specifically, one or more anomaly causes with the highest matching scores can be identified as the target environmental factor anomaly cause, such as excessively high / low temperature, insufficient / excessive humidity, insufficient / excessive light intensity, insufficient / excessive light duration, excessively high / low carbon dioxide concentration, and unreasonable substrate formula (e.g., insufficient nutrients, pH imbalance). Growth environment parameters can include one or more combinations of temperature, humidity, light intensity, carbon dioxide concentration, and substrate formula parameters. Then, based on the abnormal growth state and the target environmental factor anomaly cause, environmental parameter adjustment amounts are generated. The abnormal growth state of mushrooms can include excessively slow or excessively fast growth. Specifically: An abnormal state coefficient can be determined based on the abnormal growth state of the mushrooms; for example, a coefficient of 1.0 can be used when growth is too slow, and a coefficient of 0.8 can be used when growth is too fast. The influence weight of the abnormal environmental factor inducing the problem on the corresponding environmental parameter can be determined; for example, when the abnormal environmental factor inducing the problem is low temperature, the influence weight on the temperature parameter is 1.0, and the influence weight on the humidity parameter is 0.2. The deviation coefficient between the real-time growth rate and the normal growth rate range can be calculated. Then, based on the abnormal state coefficient, influence weight, and deviation coefficient, the environmental parameter adjustment amount can be calculated. The formula for calculating the environmental parameter adjustment amount includes:

[0098] in, Let represent the environmental parameter adjustment amount of the u-th type of environmental parameter, and Q represent the abnormal state coefficient of the fungus. This represents the weight of the influence of environmental factor anomalous causes, identified as s, on the u-th type of environmental parameter. This represents the absolute value of the environmental parameter deviation for the u-th type of environmental parameter. The deviation coefficient is represented by the following formula:

[0099] in, This indicates the real-time growth rate of the mushrooms. This represents the median of the normal growth rate range, where:

[0100] This indicates the percentage deviation of the growth rate, used to quantify the severity of the anomaly; however, this embodiment does not impose any limitations on it.

[0101] As can be seen, implementing this optional embodiment can retrieve log information of the mushroom growth environment, determine whether the abnormal causes include non-environmental factors, obtain the current growth environment parameters of the mushrooms after excluding non-environmental factors, determine the historical average values ​​of environmental parameters corresponding to the normal growth rate range and calculate the environmental parameter deviation values, match the joint feature vector and environmental parameter deviation values ​​with historical abnormal records to determine the target environmental factor abnormal cause, generate environmental parameter adjustment amounts based on the abnormal growth state and the target environmental factor abnormal cause, and adjust the growth environment. This can avoid the problem of blindly adjusting environmental parameters in traditional anomaly handling, improve the accuracy and pertinence of anomaly cause judgment, reduce the impact of ineffective operations on mushroom growth, replace manual experience judgment with scientific matching algorithms, reduce the subjectivity of cause identification, improve the reliability of anomaly handling, achieve precise control of abnormal growth state, avoid secondary damage caused by parameter mutations, help mushrooms quickly recover normal growth, reduce yield loss, improve the closed-loop management of anomaly handling, enhance the integrity and practicality of the technical solution, and further strengthen the risk prevention and control capabilities and intelligent management level of mushroom farming.

[0102] Example 3 Please see Figure 3 , Figure 3 This is a schematic diagram of a yield prediction device based on the growth status of mushrooms, as disclosed in an embodiment of the present invention. Figure 3 The described yield prediction device based on mushroom growth status can be applied in mushroom cultivation cabins. This device may include an intelligent server or intelligent platform for predicting mushroom yield. The intelligent server may include a local server or a cloud server; this embodiment of the invention is not limited to any particular type. Figure 3 As shown, the yield prediction device based on the growth status of mushrooms may include: The acquisition module 301 is used to acquire time-series data of mushrooms at preset time intervals. The time-series data includes spectral image time-series data and point cloud depth data. The first determining module 302 is used to determine the two-dimensional feature data of each mushroom fruiting body based on the time series data of the spectral image, and to determine the three-dimensional feature data of each mushroom fruiting body based on the point cloud depth data. The two-dimensional feature data includes the vegetation index, anthocyanin reflectance and fruiting body coordinates of each mushroom fruiting body, and the three-dimensional feature data includes the cap area, stipe length, stipe curvature and space occupancy rate of each mushroom fruiting body. The calculation module 303 is used to input the two-dimensional feature data and three-dimensional feature data of each mushroom fruiting body into the prediction network, calculate the real-time growth rate of the mushroom, and compare the real-time growth rate with the preset normal growth rate range to obtain the comparison result. The judgment module 304 is used to determine whether the mushroom is in the target growth stage when the comparison result indicates that the real-time growth rate is within the normal growth rate range. The prediction module 305 is used to input the real-time growth rate and the corresponding historical yield data of the mushrooms into a multivariate regression model when the mushrooms are in the target growth stage, so as to obtain the yield prediction results of the mushrooms. The first generation module 306 is used to determine that the mushroom is in an abnormal growth state when the comparison result indicates that the real-time growth rate is not within the normal growth rate range, and to generate abnormal growth alarm information.

[0103] It is evident that implementation Figure 3 The described mushroom growth state-based yield prediction device can collect time-series data of mushrooms at preset time intervals. It determines the two-dimensional feature data of each mushroom fruiting body based on spectral image time-series data and the three-dimensional feature data of each mushroom fruiting body based on point cloud depth data. The two-dimensional and three-dimensional feature data of each mushroom fruiting body are input into a prediction network to calculate the real-time growth rate of the mushroom. The real-time growth rate is compared with a preset normal growth rate range to obtain the comparison result. When the comparison result indicates that the real-time growth rate is within the normal growth rate range, it determines whether the mushroom is in the target growth stage. When the mushroom is in the target growth stage, the real-time growth rate and the corresponding historical yield data of the mushroom are input into a multivariate regression analysis. The model obtains mushroom yield prediction results. When the comparison results indicate that the real-time growth rate is not within the normal growth rate range, it determines that the mushroom is in an abnormal growth state and generates abnormal growth alarm information. It can improve the accuracy of feature recognition through refined image analysis, achieve full-dimensional coverage of two-dimensional physiological features and three-dimensional morphological features, improve the reliability and accuracy of growth status judgment and yield prediction, and realize closed-loop management of growth dynamic monitoring, early yield prediction and timely alarm of anomalies. It improves the accuracy of mushroom growth status monitoring, thereby improving the accuracy and intelligence of yield prediction. It can standardize the execution of data collection and processing processes, reduce labor costs, provide a scientific basis for large-scale breeding decisions, and help improve breeding efficiency.

[0104] In an optional embodiment, such as Figure 4 As shown, the calculation module 303 inputs the two-dimensional and three-dimensional feature data of each mushroom fruiting body into the prediction network. The specific method for calculating the real-time growth rate of the mushroom includes: The two-dimensional and three-dimensional feature data of each mushroom fruiting body are normalized to obtain the spliced ​​joint feature vector. The preset number of joint feature vectors corresponding to each mushroom fruiting body are used to generate a feature matrix in chronological order. The feature matrix is ​​then input into the prediction network to predict the growth increment of each mushroom fruiting body at the next time step. The growth increment includes the increase in cap area and the increase in stipe length. The relative growth rate of area is calculated based on the cap area and cap area increment of each mushroom fruiting body. The relative growth rate of length is calculated based on the stipe length and stipe length increment of each mushroom fruiting body. Based on the relative growth rate of area and the relative growth rate of length, the fruiting body growth rate of each mushroom fruiting body is calculated. The spatial distribution weight of each mushroom fruiting body is determined based on its spatial occupancy rate, and the real-time growth rate of the mushroom is calculated based on its fruiting body growth rate and spatial distribution weight.

[0105] It is evident that implementation Figure 4The described mushroom growth state-based yield prediction device can collect time-series data of mushrooms at preset time intervals. It determines the two-dimensional feature data of each mushroom fruiting body based on spectral image time-series data and the three-dimensional feature data of each fruiting body based on point cloud depth data. The device normalizes the two-dimensional and three-dimensional feature data of each fruiting body to obtain a spliced ​​joint feature vector. A preset number of joint feature vectors corresponding to each fruiting body are then used to generate a feature matrix in chronological order and input into a prediction network to predict growth increments. The relative growth rates of area and length are calculated based on cap area, stipe length, and corresponding increments to obtain the fruiting body growth rate. Combined with spatial distribution weights determined by spatial occupancy rate, the device calculates the real-time growth rate of the mushroom. This eliminates interference from features of different dimensions, ensures the data consistency and reliability of the joint feature vectors, accurately captures the dynamic changes in mushroom growth, takes into account the growth characteristics of different parts of the mushroom and the differences in spatial resource occupancy of fruiting bodies, and avoids misjudgments of growth rate caused by single indicators or unweighted calculations. The standardized and refined calculation process improves the accuracy and scientific rigor of real-time growth rate calculations. The system inputs 2D and 3D feature data into a prediction network to calculate the real-time growth rate of mushrooms. This real-time growth rate is then compared with a preset normal growth rate range. If the comparison result indicates that the real-time growth rate is within the normal range, it determines whether the mushroom is in the target growth stage. If the mushroom is in the target growth stage, the real-time growth rate and corresponding historical yield data are input into a multivariate regression model to obtain the mushroom yield prediction result. If the comparison result indicates that the real-time growth rate is outside the normal range, it determines that the mushroom is in an abnormal growth state and generates an abnormal growth alarm message. This system can improve feature recognition accuracy through refined image analysis, achieving full-dimensional coverage of 2D physiological features and 3D morphological features, thus improving the reliability and accuracy of growth state judgment and yield prediction. It enables closed-loop management of dynamic growth monitoring, early yield prediction, and timely anomaly alarms, improving the accuracy of mushroom growth state monitoring, thereby increasing the accuracy and intelligence of yield prediction. It also standardizes data collection and processing procedures, reduces labor costs, provides a scientific basis for large-scale farming decisions, and helps improve farming efficiency.

[0106] In another alternative embodiment, such as Figure 4 As shown, the prediction module 305 inputs the real-time growth rate and the corresponding historical yield data of mushrooms into the multivariate regression model to obtain the mushroom yield prediction results. The specific methods include: Filter the target historical yield data from the historical yield data of mushrooms. The deviation between the historical growth environment parameters corresponding to the target historical yield data and the current growth environment parameters of mushrooms is less than the preset deviation threshold. Extract the historical growth rate, historical joint feature vector, and historical yield benchmark value of mushrooms from the target historical yield data. Input the joint feature vector, real-time growth rate, historical growth rate, historical joint feature vector, and historical yield benchmark value of mushrooms into a multivariate regression model to obtain the predicted value of each mushroom fruiting body. The predicted value of the fruiting body includes the predicted value of the weight of a single fruiting body and the predicted value of the number of fruiting bodies. Based on historical yield data analysis, the average historical prediction error rate of the multivariate regression model is analyzed, and the yield prediction results of mushrooms are calculated based on the average historical prediction error rate and the predicted value of each mushroom fruiting body.

[0107] It is evident that implementation Figure 4 The described mushroom growth status-based yield prediction device can filter target historical yield data from historical mushroom yield data, identifying data where the deviation between historical and current growth environment parameters is less than a preset threshold. It extracts historical growth rate, historical joint feature vector, and historical yield benchmark value from this target historical yield data. The device then inputs these parameters into a multivariate regression model to obtain predicted values ​​for the mushrooms. Combining this with the historical prediction average error rate, the device calculates the mushroom yield prediction result. This ensures a high degree of adaptability between historical data and the current cultivation environment, avoiding interference from irrelevant historical data. It integrates multi-dimensional core features as model input, improving the scientific rigor and comprehensiveness of yield prediction. Furthermore, it corrects preliminary prediction results using the historical prediction average error rate, effectively offsetting inherent model biases and significantly improving the accuracy and reliability of yield prediction. Finally, it outputs detailed predicted values ​​for the weight and quantity of individual mushrooms, expanding the application scenarios of the prediction results and providing data support for refined management such as harvesting arrangements and graded sales, further enhancing the scientific basis of mushroom cultivation decisions.

[0108] In yet another alternative embodiment, such as Figure 4 As shown, the yield prediction device based on the growth status of mushrooms may further include: The first acquisition module 307 is used to acquire sales data of various types of mushrooms in the target sales area corresponding to the mushrooms within a historical time period. The sales data includes price fluctuation data of each type of mushroom within the historical time period. The prediction module 305 is also used to predict the market sales forecast data of mushrooms at the target harvest stage based on the sales data of each type of mushroom. The market sales forecast data includes the predicted unit price, transportation unit price and packaging material unit price. The first acquisition module 307 is also used to acquire mushroom cultivation cost data, and input the market sales forecast data and cultivation cost data of mushrooms at the target harvest stage into the preset first profit maximization planning model to obtain the predicted optimal profit data for mushrooms. The predicted optimal profit data includes the optimal harvest time, the optimal harvest quantity, and the predicted profit value.

[0109] It is evident that implementation Figure 4 The described yield prediction device based on mushroom growth status can acquire sales data of various mushroom varieties in the target sales area over a historical period. Based on the sales data of each mushroom variety, it predicts market sales data for the target harvest stage, acquires mushroom cultivation cost data, and inputs the market sales forecast data and cultivation cost data into a preset first-maximum-profit planning model to obtain the optimal predicted profit data for the mushrooms. This enables synergistic optimization of yield and profit, improving the economic benefits of mushroom farming. It can accurately capture dynamic market price trends, avoiding decision-making biases caused by fixed-price estimations, and improving the reliability and adaptability of profit prediction. It can transform abstract profit analysis into directly applicable harvesting decision guidelines, solving the problem of the disconnect between profit analysis and actual farming operations. It covers the core decision-making needs of the entire farming cycle, further enhancing the intelligence and marketization level of mushroom farming and helping farmers maximize their profits.

[0110] In yet another alternative embodiment, such as Figure 4 As shown, the first acquisition module 307 is also used to acquire the planting space data of the mushroom planting site and the maximum planting area limit for each type of mushroom in the planting site. The planting space data includes the maximum effective planting area. The yield prediction device based on the growth status of mushrooms may also include: The second determining module 308 is also used to determine the unit yield forecast profit of each type of mushroom based on the sales data of each type of mushroom, and to determine the unit area yield standard value of each type of mushroom based on historical yield data. The planning module 309 is used to input planting space data, the maximum planting area limit for each type of mushroom, the predicted profit per unit yield of each type of mushroom, and the standard value of the yield per unit area for each type of mushroom into a preset second profit-maximizing planning model to obtain the optimal planting data for each type of mushroom. The optimal planting data includes the planting area for each type of mushroom. The prediction module 305 is also used to predict the predicted price trend of each type of mushroom in the target sales area within a future preset time period based on the sales data of various types of mushrooms in the target sales area during a historical period, and to optimize the optimal planting data of each type of mushroom based on the predicted price trend of each type of mushroom. The second generation module 310 is used to generate a planting plan report based on the optimized optimal planting data and the growth cycle data of each type of mushroom. The planting plan report includes the planting area and batch planting time of each type of mushroom.

[0111] It is evident that implementation Figure 4 The described yield prediction device based on mushroom growth status can acquire planting space data of mushroom cultivation sites and the maximum planting area limit for each mushroom variety. It determines the predicted profit per unit yield based on sales data for each mushroom variety and the standard yield per unit area based on historical yield data. The planting space data, maximum planting area limit, predicted profit per unit yield, and standard yield per unit area are input into a second profit-maximizing planning model to obtain the optimal planting data for each mushroom variety. This optimal planting data is further optimized by combining the predicted price trend of each mushroom variety in the target sales area over a preset future time period. Based on the optimized optimal planting data and... The growth cycle data of each type of mushroom generates a planting plan report, which enables the scientific optimization of planting space and mushroom varieties, avoiding resource waste and the risk of over-planting of a single variety, improving space utilization, dynamically adapting to market price fluctuations, enhancing the risk resistance and market adaptability of the planting plan, avoiding price drops caused by concentrated market entry by planting in batches and staggering harvesting times, ensuring stable breeding income, generating standardized and structured planting plan reports, reducing the difficulty of breeding decisions, and further improving the stability of large-scale breeding benefits and the level of intelligent management by reasonably combining multiple varieties to diversify market risks.

[0112] In yet another alternative embodiment, such as Figure 4 As shown, the yield prediction device based on the growth status of mushrooms may further include: The retrieval module 311 is used to retrieve the log information of the mushroom's growth environment when the comparison result indicates that the real-time growth rate is not within the normal growth rate range, and to determine whether the abnormal factors causing the mushroom to be in an abnormal growth state include non-environmental abnormal factors based on the log information. The log information includes personnel operation information and inspection information. The second acquisition module 312 is used to acquire the current growth environment parameters of the mushroom when it is determined that the abnormal cause does not include non-environmental abnormal causes. The current growth environment parameters include at least one of temperature parameters, humidity parameters, light parameters, carbon dioxide concentration parameters, and substrate formula parameters. The first determining module 302 is also used to determine the historical average value of environmental parameters corresponding to the normal growth rate range, and to calculate the environmental parameter deviation value based on the current growth environmental parameters and the historical average value of environmental parameters; The first determining module 302 is also used to match the joint feature vector and the environmental parameter deviation value with historical anomaly records, determine the matching degree score corresponding to each environmental factor anomaly cause, and determine the target environmental factor anomaly cause based on the matching degree score corresponding to each environmental factor anomaly cause. The first generation module 306 is also used to generate environmental parameter adjustment amounts based on abnormal growth status and abnormal inducing factors of target environmental factors, and to adjust the growth environment of mushrooms based on the environmental parameter adjustment amounts.

[0113] It is evident that implementation Figure 4 The described yield prediction device based on mushroom growth status can retrieve log information of the mushroom growth environment, determine whether the abnormal causes include non-environmental factors, obtain the current growth environment parameters of the mushrooms after excluding non-environmental factors, determine the historical average values ​​of environmental parameters corresponding to the normal growth rate range and calculate the environmental parameter deviation values, match the joint feature vector and environmental parameter deviation values ​​with historical abnormal records to determine the target environmental factor abnormal cause, generate environmental parameter adjustment amounts based on the abnormal growth status and target environmental factor abnormal cause, and adjust the growth environment accordingly. This avoids the problem of blindly adjusting environmental parameters in traditional anomaly handling, improves the accuracy and targeting of anomaly cause judgment, reduces the impact of ineffective operations on mushroom growth, and can replace manual experience judgment with scientific matching algorithms, reducing the subjectivity of cause identification, improving the reliability of anomaly handling, achieving precise control of abnormal growth status, avoiding secondary damage caused by parameter mutations, helping mushrooms quickly recover normal growth, reducing yield loss, improving the closed-loop management of anomaly handling, enhancing the integrity and practicality of the technical solution, and further strengthening the risk prevention and control capabilities and intelligent management level of mushroom farming.

[0114] In yet another alternative embodiment, such as Figure 4 As shown, the acquisition module 301 is also used to acquire full data of the cultivation cycle after completing the cultivation cycle of the mushrooms, and store the full data in a preset training database. The full data includes the two-dimensional feature data, the three-dimensional feature data, the growth increment, the real-time growth rate, the yield prediction result, the measured growth increment, the measured growth rate, and the actual harvest yield. The calculation module 303 is also used to calculate the growth increment prediction error of the prediction network and the yield prediction error of the multivariate regression model in each full dataset when the number of full datasets in the training database reaches the preset training trigger number. The yield prediction device based on the growth status of mushrooms may also include: Training module 313 is used to incrementally train the prediction network and multivariate regression model using each full dataset in the training database and the corresponding growth increment prediction error and yield prediction error, so as to obtain the average growth increment prediction error value of the prediction network and the average yield prediction error value of the multivariate regression model. The judgment module 304 is also used to judge whether the average error value of the growth increment prediction and the average error value of the yield prediction meet the preset training accuracy requirements. When the average error value of the growth increment prediction and the average error value of the yield prediction meet the training accuracy requirements, the model optimization information of the trained prediction network and the multivariate regression model is determined. The model optimization information includes parameter configuration information, error optimization magnitude and mushroom species adaptation information. The filtering module 314 is used to store the trained prediction network and multivariate regression model in the model version management library. When the next batch of mushroom cultivation operation is started, the matching target prediction model and target multivariate regression model are filtered in the model version management library according to the mushroom variety of the batch of mushrooms, so as to perform the real-time growth rate prediction operation and yield prediction operation of the batch of mushrooms.

[0115] It is evident that implementation Figure 4 The described yield prediction device based on mushroom growth status can collect all data for the mushroom cultivation cycle after completion and store it in a preset training database. When the amount of data in the training database reaches the preset training trigger number, the growth rate prediction error of the prediction network and the yield prediction error of the multivariate regression model in each data set are calculated. The two models are incrementally trained using the data in the training database and the corresponding prediction errors. When the training accuracy requirements are met, the model optimization information is determined and stored in the model version management library. In the next batch of cultivation, the target model is selected and matched according to the mushroom category to perform the prediction operation. It can continuously supplement new data through incremental training, solve the problem of the decline in prediction accuracy of traditional models after long-term use, and achieve continuous improvement of model accuracy. It can specifically correct model biases, improve optimization efficiency and the extent of improvement in prediction accuracy. It can achieve accurate adaptation of the model to different mushroom categories and cultivation scenarios, avoid prediction distortion caused by cross-scenario use, improve the reusability and implementation efficiency of the technical solution, reduce the labor cost of model maintenance, adapt to new situations such as changes in the cultivation environment and mushroom variety improvement, and ensure the long-term stable accuracy of the entire prediction method, providing reliable technical support for the continuous intelligent upgrading of mushroom cultivation.

[0116] Example 4 Please see Figure 5 , Figure 5 This is a schematic diagram of another yield prediction device based on the growth status of mushrooms disclosed in an embodiment of the present invention. Figure 5As shown, the yield prediction device based on the growth status of mushrooms may include: Memory 401 storing executable program code; Processor 402 coupled to memory 401; The processor 402 calls the executable program code stored in the memory 401 to execute the steps in the yield prediction method based on the growth status of mushrooms as described in Embodiment 1 or Embodiment 2 of the present invention.

[0117] Example 5 This invention discloses a computer storage medium storing computer instructions. When these computer instructions are invoked, they are used to execute some or all of the steps in any of the yield prediction methods based on mushroom growth status disclosed in Embodiment 1 of this invention.

[0118] Example 6 This invention discloses a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program, and the computer program is operable to cause a computer to perform the steps in the yield prediction method based on mushroom growth status described in Embodiment 1 or Embodiment 2.

[0119] The device embodiments described above are merely illustrative. The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; that is, they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0120] Through the detailed description of the above embodiments, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, including read-only memory (ROM), random access memory (RAM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), one-time programmable read-only memory (OTPROM), electrically-Erasable Programmable Read-Only Memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, disk storage, magnetic tape storage, or any other computer-readable medium that can be used to carry or store data.

[0121] Finally, it should be noted that the yield prediction method and apparatus based on the growth status of mushrooms disclosed in the embodiments of the present invention are merely preferred embodiments of the present invention and are only used to illustrate the technical solutions of the present invention, not to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for yield prediction based on the growth state of a mushroom, characterized by, The method comprises: Collecting time series data of the mushrooms at preset time intervals, the time series data comprising spectral image time series data and point cloud depth data; Determining two-dimensional feature data of each mushroom entity from the spectral image time series data, the two-dimensional feature data comprising vegetation index, anthocyanin reflectance and entity coordinate of each mushroom entity; and determining three-dimensional feature data of each mushroom entity from the point cloud depth data, the three-dimensional feature data comprising cap area, stem length, stem curvature and space occupancy rate of each mushroom entity; Inputting the two-dimensional feature data and the three-dimensional feature data of each mushroom entity into a prediction network, calculating a real-time growth rate of the mushrooms, and comparing the real-time growth rate with a preset normal growth rate range to obtain a comparison result; When the comparison result indicates that the real-time growth rate is within the normal growth rate range, determining whether the mushrooms are in a target growth stage; when the mushrooms are in the target growth stage, inputting the real-time growth rate and historical yield data corresponding to the mushrooms into a multivariate regression model to obtain yield prediction results of the mushrooms; When the comparison result indicates that the real-time growth rate is not within the normal growth rate range, determining that the mushrooms are in an abnormal growth state, and generating growth abnormality alarm information of the abnormal growth state.

2. The yield prediction method based on the growth state of mushrooms according to claim 1, characterized in that, The inputting of the two-dimensional feature data and the three-dimensional feature data of each mushroom entity into the prediction network, and the calculation of the real-time growth rate of the mushrooms, comprises: Normalizing the two-dimensional feature data and the three-dimensional feature data of each mushroom entity to obtain a spliced joint feature vector; Generating a feature matrix in time sequence from a preset number of joint feature vectors corresponding to each mushroom entity, and inputting the feature matrix into the prediction network to predict a growth increment of each mushroom entity at the next time, the growth increment comprising cap area increment and stem length increment; Calculating an area relative growth rate of each mushroom entity from the cap area and the cap area increment, calculating a length relative growth rate of each mushroom entity from the stem length and the stem length increment, and calculating a entity growth rate of each mushroom entity from the area relative growth rate and the length relative growth rate; Determining a space distribution weight of each mushroom entity from the space occupancy rate of each mushroom entity, and calculating a real-time growth rate of the mushrooms from the entity growth rate and the space distribution weight of each mushroom entity.

3. The yield prediction method based on the growth state of mushrooms according to claim 2, characterized in that, The inputting of the real-time growth rate and the historical yield data corresponding to the mushrooms into the multivariate regression model, and the obtaining of the yield prediction results of the mushrooms, comprises: Filtering target historical yield data from the historical yield data corresponding to the mushrooms, the environmental parameter deviation between a historical growth environment parameter corresponding to the target historical yield data and a current growth environment parameter of the mushrooms being less than a preset deviation threshold. extracting a historical growth rate, a historical joint feature vector and a historical yield benchmark value of the mushroom corresponding to the target historical yield data, and inputting the joint feature vector of the mushroom, the real-time growth rate, the historical growth rate, the historical joint feature vector and the historical yield benchmark value into the multivariate regression model to obtain a sub-entity predicted value of each of the mushroom entities, the sub-entity predicted value including a single sub-entity weight predicted value and a sub-entity number predicted value; analyzing a historical prediction average error rate of the multivariate regression model according to the historical yield data, and calculating a yield prediction result of the mushroom according to the historical prediction average error rate and the sub-entity predicted value of each of the mushroom entities.

4. The yield prediction method based on the growth state of mushrooms according to any one of claims 1 to 3, characterized in that, The method further comprises: obtaining sales data of multiple mushroom categories in a target sales area corresponding to the mushroom in a historical time period, the sales data including price fluctuation data of each of the mushroom categories in the historical time period; predicting market sales prediction data of the mushroom in a target harvesting stage according to the sales data of each of the mushroom categories, the market sales prediction data including a predicted unit price, a transportation unit price and a packaging consumable unit price; obtaining planting cost data of the mushroom, and inputting the market sales prediction data of the mushroom in the target harvesting stage and the planting cost data into a preset first maximum yield planning model to obtain predicted optimal yield data of the mushroom, the predicted optimal yield data including an optimal harvesting time, an optimal harvesting quantity and a predicted yield value.

5. The yield prediction method based on the growth state of mushrooms according to claim 4, characterized in that, The method further comprises: obtaining planting space data of a planting site of the mushroom and a maximum planting area limit of each of the mushroom categories in the planting site, the planting space data including a maximum effective planting area; determining a unit yield predicted profit of each of the mushrooms according to the sales data of each of the mushroom categories, and determining a unit area yield standard value of each of the mushroom categories according to the historical yield data; inputting the planting space data, the maximum planting area limit of each of the mushroom categories, the unit yield predicted profit of each of the mushrooms and the unit area yield standard value of each of the mushroom categories into a preset second maximum yield planning model to obtain optimal planting data of each of the mushroom categories, wherein the optimal planting data includes a planting area of each of the mushroom categories; predicting a predicted price trend of each of the mushroom categories in a future preset time period in the target sales area according to the sales data of multiple mushroom categories in the target sales area in a historical time period, and optimizing the optimal planting data of each of the mushroom categories according to the predicted price trend of each of the mushroom categories; generating a planting plan report according to the optimized optimal planting data and growth cycle data of each of the mushroom categories, the planting plan report including a planting area and a batch planting time of each of the mushroom categories.

6. The yield prediction method based on the growth state of mushrooms according to claim 2 or 3, characterized by, When the comparison result indicates that the real-time growth rate is not within the normal growth rate range, the method further comprises: obtain log information of a growth environment of the mushroom, and determine whether an abnormal inducement causing the mushroom to be in the abnormal growth state includes a non-environmental factor abnormal inducement according to the log information, the log information including personnel operation information and inspection information; when it is determined that the abnormal inducement does not include the non-environmental factor abnormal inducement, obtain a current growth environment parameter of the mushroom, the current growth environment parameter including at least one of a temperature parameter, a humidity parameter, an illumination parameter, a carbon dioxide concentration parameter, and a substrate formula parameter; determine a historical average value of an environmental parameter corresponding to the normal growth rate range, and calculate an environmental parameter deviation value according to the current growth environment parameter and the historical average value of the environmental parameter; match the joint feature vector and the environmental parameter deviation value with historical abnormal records to determine a matching degree score corresponding to each environmental factor abnormal inducement, and determine a target environmental factor abnormal inducement according to the matching degree score corresponding to each environmental factor abnormal inducement; generate an environmental parameter adjustment amount according to the abnormal growth state and the target environmental factor abnormal inducement, and adjust the growth environment of the mushroom according to the environmental parameter adjustment amount.

7. The yield prediction method based on the growth state of mushrooms according to any one of claims 1 to 3, characterized in that, The method further includes: after completing a cultivation period for the mushroom, collect full-amount data of the cultivation period, and store the full-amount data into a preset training database, the full-amount data including the two-dimensional feature data, the three-dimensional feature data, the growth increment, the real-time growth rate, the yield prediction result, a measured growth increment, a measured growth rate, and an actual harvested yield; when a number of the full-amount data in the training database reaches a preset training trigger number, calculate a growth increment prediction error of the prediction network and a yield prediction error of the multivariate regression model in each of the full-amount data; perform incremental training on the prediction network and the multivariate regression model through each of the full-amount data in the training database and the growth increment prediction error and the yield prediction error corresponding to each of the full-amount data, to obtain a growth increment prediction average error value corresponding to the prediction network and a yield prediction average error value corresponding to the multivariate regression model; determine whether the growth increment prediction average error value and the yield prediction average error value meet a preset training accuracy requirement, and when the growth increment prediction average error value and the yield prediction average error value meet the training accuracy requirement, determine model optimization information of the trained prediction network and the multivariate regression model, the model optimization information including parameter configuration information, error optimization amplitude, and mushroom species adaptation information; store the trained prediction network and the multivariate regression model into a model version management library, and when starting a next batch of mushroom cultivation operation, filter a target prediction model and a target multivariate regression model matching a mushroom species of the batch of mushrooms in the model version management library, to perform real-time growth rate prediction operation and yield prediction operation on the batch of mushrooms.

8. A yield prediction device based on the growth status of mushrooms, characterized in that, The device includes: The collection module is used for collecting time sequence data of the mushroom according to a preset time interval, and the time sequence data includes spectral image time sequence data and point cloud depth data; The first determination module is used for determining two-dimensional feature data of each mushroom entity according to the spectral image time sequence data, and determining three-dimensional feature data of each mushroom entity according to the point cloud depth data, wherein the two-dimensional feature data includes vegetation index, anthocyanin reflectivity and entity coordinate of each mushroom entity, and the three-dimensional feature data includes cap area, stem length, stem bending rate and space occupancy rate of each mushroom entity; The calculation module is used for inputting the two-dimensional feature data and the three-dimensional feature data of each mushroom entity into a prediction network, calculating a real-time growth rate of the mushroom, and comparing the real-time growth rate with a preset normal growth rate range to obtain a comparison result; The judgment module is used for judging whether the mushroom is in a target growth stage when the comparison result indicates that the real-time growth rate is within the normal growth rate range; The prediction module is used for inputting the real-time growth rate and historical yield data corresponding to the mushroom into a multivariate regression model to obtain a yield prediction result of the mushroom when the mushroom is in the target growth stage; The first generation module is used for determining that the mushroom is in an abnormal growth state and generating growth abnormality alarm information of the abnormal growth state when the comparison result indicates that the real-time growth rate is not within the normal growth rate range.

9. A yield prediction device based on the growth status of mushrooms, characterized in that, The device comprises: a memory storing executable program codes; a processor coupled with the memory; the processor invokes the executable program codes stored in the memory to execute the yield prediction method based on the mushroom growth state according to any one of claims 1-7.

10. A computer storage medium, characterized in that, The computer storage medium stores computer instructions, and the computer instructions are used to execute the yield prediction method based on the mushroom growth state according to any one of claims 1-7 when invoked. The computer storage medium stores computer instructions, and the computer instructions are used to execute the yield prediction method based on the mushroom growth state according to any one of claims 1-7 when invoked.

Images