Method for determining optimal sampling time of seedless grape embryo rescue fruits

Abstract

The invention provides a seedless grape embryo rescue fruit optimal sampling time determination method, which comprises the following steps: determining a target variety of a to-be-sampled fruit, and obtaining a corresponding target hardness interval according to the target variety; acquiring a near-infrared spectral reflectivity curve of the to-be-sampled fruit in real time every first preset duration, and obtaining the real-time peel hardness of the to-be-sampled fruit according to the near-infrared spectral reflectivity curve; and if the real-time peel hardness is within the target hardness interval, sampling the to-be-sampled fruit. According to the technical scheme, the influence of subjective factors can be reduced, the sampling of seedless grape embryo rescue fruits is nondestructively and accurately guided for the first time, so that the success rate of embryo rescue is improved, the corresponding target hardness interval can be determined according to the target variety, and personalized sampling time determination is realized. The method has wide applicability to embryo rescue work of various grape varieties, and opens up a new view for accelerating the breeding process of new seedless grape varieties.

Description

Technical Field

[0001] The present application relates to the technical field of seedless grape embryo rescue fruit sampling, and in particular to a method for determining an optimal sampling time for seedless grape embryo rescue fruit. Background Art

[0002] Embryo rescue technology plays a key role in seedless grape hybrid breeding and genetic improvement. Since some seedless grape varieties cannot be passed down due to incomplete embryo development or hybrid incompatibility under natural conditions, embryo rescue technology can be used to remove ovules before embryo abortion, and artificial tissue culture can be performed to obtain hybrid seedlings, thereby obtaining new and excellent seedless grape varieties. However, the success rate of embryo rescue depends largely on the accuracy of sampling time. If sampling is too early, the embryo may not develop completely and it is difficult to continue to grow under culture conditions; if sampling is too late, the embryo has already aborted, which also reduces the success rate of rescue.

[0003] In the prior art, the method for determining the sampling time of seedless grape embryo rescue fruit mainly relies on empirical observation. For example, the embryo development stage is judged by observing the appearance changes of the fruit (such as color, size, etc.), but this method is highly subjective and has low accuracy. Summary of the invention

[0004] In view of the above defects or deficiencies in the prior art, the present application aims to provide a method for determining the optimal sampling time for seedless grape embryo rescue fruit, comprising the following steps: Determine the target variety of the fruit to be sampled, and obtain the corresponding target hardness range according to the target variety; At intervals of a first preset time length, a near-infrared spectral reflectance curve of the fruit to be sampled is obtained in real time, and the real-time peel hardness of the fruit to be sampled is obtained according to the near-infrared spectral reflectance curve; If the real-time fruit skin hardness is within the target hardness range, the fruit to be sampled is sampled.

[0005] According to the technical solution provided in the embodiment of the present application, after obtaining the real-time peel hardness of the fruit to be sampled according to the near-infrared spectral reflectance curve, the following steps are also included: The real-time peel hardness is stored in a historical data set corresponding to the moment before the current moment to obtain the historical data set corresponding to the current moment, wherein the historical data set includes several groups of the real-time peel hardness and the acquisition time corresponding to each group of the real-time peel hardness.

[0006] According to the technical solution provided in the embodiment of the present application, after obtaining the real-time peel hardness of the fruit to be sampled according to the near-infrared spectral reflectance curve, the following steps are also included: If the real-time peel hardness does not reach the target hardness interval, determining whether the real-time peel hardness reaches a first preset threshold, the first preset threshold being less than a lower limit of the target hardness interval; If the real-time peel hardness reaches the first preset threshold, the development rate is retrieved and obtained based on the historical data set; According to the growth rate, a predicted sampling time is obtained, and then a first interval length between the current time and the predicted sampling time is obtained; If the first interval duration is less than or equal to the second preset duration, the first preset duration is shortened according to the first interval duration.

[0007] According to the technical solution provided in the embodiment of the present application, obtaining the real-time peel hardness of the fruit to be sampled according to the near-infrared spectral reflectance curve specifically includes the following steps: According to the target variety, a target recognition model corresponding to the target variety is obtained; The near-infrared spectral reflectance curve is input into the target recognition model to obtain the real-time peel hardness.

[0008] According to the technical solution provided in the embodiment of the present application, obtaining the corresponding target hardness range according to the target variety specifically includes the following steps: Retrieving and traversing a model database, wherein the model database includes multiple types of fruit varieties and a prediction model corresponding to each type of the fruit variety; the prediction model characterizes the relationship between the fruit skin hardness and the embryo development rate; If the model database contains the fruit variety corresponding to the target variety, the fruit skin hardness corresponding to the maximum embryo development rate is obtained through the prediction model; The peel hardness corresponding to the maximum embryo development rate is taken as the middle value of the target hardness range corresponding to the target variety, thereby obtaining the target hardness range.

[0009] According to the technical solution provided in the embodiment of the present application, after retrieving and traversing the model database, the following steps are also included: If the model database does not contain the fruit variety corresponding to the target variety, the corresponding target hardness range is obtained through a universal recognition model.

[0010] According to the technical solution provided in the embodiment of the present application, if the model database does not contain the fruit variety corresponding to the target variety, after obtaining the corresponding target hardness range through the universal recognition model, the following steps are also included: According to the middle value of the target hardness range obtained by the general recognition model and the corresponding embryo development rate, an initial model is trained to obtain a specific prediction model; The specific prediction model corresponding to the target variety is stored in the model database.

[0011] According to the technical solution provided in the embodiment of the present application, after taking the peel hardness corresponding to the maximum embryo development rate as the middle value of the target hardness range corresponding to the target variety, the following steps are also included: According to the target variety, a first preset range is obtained; The step of obtaining the target hardness range specifically comprises the following steps: The target hardness interval is taken as the target hardness interval.

[0012] According to the technical solution provided in the embodiment of the present application, obtaining the first preset range according to the target variety specifically includes the following steps: Obtaining a target hardness change rate corresponding to the target variety; The first preset range is calculated according to the target hardness change rate.

[0013] According to the technical solution provided in the embodiment of the present application, obtaining the target hardness change rate corresponding to the target variety specifically includes the following steps: Retrieving multiple real-time peel hardnesses of the target variety and the corresponding acquisition times from the historical database to obtain a reference database; the multiple real-time peel hardnesses in the reference database cover multiple growth stages of the fruit to be sampled; Based on the reference database, obtaining hardness change rates in multiple stages; The target hardness change rate is obtained by weighted averaging the hardness change rates of each stage, wherein the harderness change rate of the stage closer to the current growth stage is given a higher weight.

[0014] Compared with the prior art, the present application has the following beneficial effects: the present application determines the target hardness interval corresponding to the target variety and obtains the near-infrared spectral reflectance curve in real time and non-destructively to obtain the real-time skin hardness, which can more accurately judge the development state of the fruit and the maturity of the embryo. Compared with the traditional method, the technical solution can reduce the influence of subjective factors, improve the accuracy of the sampling time, and for the first time achieve non-destructive and accurate guidance of the sampling of seedless grape embryo rescue fruit, thereby improving the success rate of embryo rescue. Different varieties of grape fruits have different development laws and hardness change characteristics. The technical solution can determine the corresponding target hardness interval according to the target variety and realize personalized sampling time determination. This has wide applicability to the embryo rescue work of various grape varieties and can meet the special needs of different varieties. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 A flowchart of the steps of the method for determining the optimal sampling time for seedless grape embryo rescue fruit provided in an embodiment of the present application; Figure 2 A schematic diagram of the relationship between peel hardness and embryo development rate provided in the examples of the present application. DETAILED DESCRIPTION

[0016] The present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It is to be understood that the specific embodiments described herein are only used to explain the relevant invention, rather than to limit the invention. It is also necessary to explain that, for ease of description, only the parts related to the invention are shown in the accompanying drawings.

[0017] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0018] Example 1 As mentioned in the background technology, in view of the problems in the prior art, this application proposes a method for determining the optimal sampling time for seedless grape embryos to save fruits, please refer to Figure 1 As shown, the following steps are included: S1. Determine the target variety of the fruit to be sampled, and obtain the corresponding target hardness range according to the target variety; For example, for variety A, when the fruit skin hardness is within the range of [X1, X2], the success rate of embryo rescue is the highest, and this range is the target hardness range of variety A.

[0019] S2, acquiring in real time a near-infrared spectral reflectance curve of the fruit to be sampled at intervals of a first preset time, and obtaining the real-time peel hardness of the fruit to be sampled according to the near-infrared spectral reflectance curve; Specifically, a near-infrared spectrometer is installed in the grape planting area to ensure that the instrument can accurately measure the near-infrared spectral reflectance curve of the fruit to be sampled. A portable or fixed spectrometer can be selected and installed and debugged according to actual conditions. Determine the first preset duration, that is, the time interval for measuring the near-infrared spectral reflectance curve. This time interval can be determined based on the growth rate of the fruit and the speed of embryo development. For faster-growing varieties, the time interval can be shortened to grasp the development of the fruit more promptly; for slower-growing varieties, the time interval can be appropriately extended to reduce the workload of measurement. Exemplarily, the first preset duration is measured once every 24 hours.

[0020] S3. If the real-time peel hardness is within the target hardness range, sampling the fruit to be sampled.

[0021] Specifically, sampling can be done manually or automatically to ensure the accuracy and completeness of the sampling. During the sampling process, care should be taken to avoid damaging the fruit to avoid affecting the development of the embryo and the subsequent culture process. At the same time, the sampling time, fruit variety, real-time peel hardness and other information should be recorded for subsequent analysis and research.

[0022] In a preferred embodiment, after obtaining the real-time peel hardness of the fruit to be sampled according to the near-infrared spectral reflectance curve, the method further comprises the following steps: The real-time peel hardness is stored in a historical data set corresponding to the moment before the current moment to obtain the historical data set corresponding to the current moment, wherein the historical data set includes several groups of the real-time peel hardness and the acquisition time corresponding to each group of the real-time peel hardness.

[0023] Exemplarily, the real-time skin hardness of the peel is obtained once every 24 hours. If the real-time skin hardness obtained for the Nth time has not reached the target hardness range, the historical data set corresponding to the N-1th time is retrieved. The number of groups of real-time skin hardness in the historical data set corresponding to the N-1th time is N-1 groups. The real-time skin hardness obtained for the Nth time is stored in the historical data set corresponding to the N-1th time, that is, the historical data set corresponding to the Nth time is obtained. The number of groups of real-time skin hardness in the historical data set corresponding to the N-1th time is N groups.

[0024] Specifically, by using the hardness-time relationship in the historical data set and combining it with appropriate mathematical models (such as linear regression, polynomial fitting, etc.), it is possible to predict future changes in peel hardness. For example, by linear fitting the previous data, it is found that the peel hardness is increasing at a certain rate. Based on this, it can be inferred whether the peel hardness will enter the target hardness range in the next few days, so as to prepare for sampling in advance.

[0025] In a preferred embodiment, after obtaining the real-time peel hardness of the fruit to be sampled according to the near-infrared spectral reflectance curve, the method further comprises the following steps: If the real-time peel hardness does not reach the target hardness interval, determining whether the real-time peel hardness reaches a first preset threshold, the first preset threshold being less than a lower limit of the target hardness interval; Specifically, according to the characteristics of the target variety and previous experimental data, a first preset threshold value that is less than the lower limit of the target hardness range is determined. For example, for a certain grape variety, the target hardness range is [600g, 700g], and after analysis, the first preset threshold value is determined to be 500g.

[0026] If the real-time peel hardness reaches the first preset threshold, the development rate is retrieved and obtained based on the historical data set; According to the growth rate, a predicted sampling time is obtained, and then a first interval length between the current time and the predicted sampling time is obtained; For example, if the current peel hardness is 350g, the development rate is 5g per hour, and the lower limit of the target hardness range is 400g, then the time required to reach the target hardness range can be calculated as (400g - 350g) ÷ 5g / hour = 10 hours, that is, the predicted sampling time is the current time plus 10 hours.

[0027] If the first interval duration is less than or equal to the second preset duration, the first preset duration is shortened according to the first interval duration.

[0028] If the first interval duration is less than or equal to the second preset duration, the first preset duration is shortened according to the first interval duration.

[0029] Specifically, if the first interval is less than or equal to the second preset time, it means that the fruit may soon enter the target hardness range, and the first preset time needs to be shortened to monitor the change in skin hardness more frequently. For example, if the first preset time is originally 24 hours, the second preset time is 12 hours, and the calculated first interval is 10 hours, which is less than the second preset time, the first preset time can be shortened to a time interval of 12 hours or less.

[0030] This embodiment can dynamically adjust the first preset duration according to the development of the fruit, and can increase the monitoring frequency when the fruit is close to the sampling time. In this way, the changes in the hardness of the fruit skin can be more accurately grasped, ensuring that sampling is performed at the best time, and improving the success rate of embryo rescue. At the same time, when the fruit is far away from the target hardness range, monitoring is performed according to the longer first preset duration, saving manpower, material resources and time costs. When the fruit is close to the sampling time, the monitoring frequency can be adjusted in time, improving the utilization efficiency of resources.

[0031] In a preferred embodiment, obtaining the real-time peel hardness of the fruit to be sampled according to the near-infrared spectral reflectance curve specifically comprises the following steps: According to the target variety, a target recognition model corresponding to the target variety is obtained; The near-infrared spectral reflectance curve is input into the target recognition model to obtain the real-time peel hardness.

[0032] In a preferred embodiment, obtaining the corresponding target hardness range according to the target variety specifically includes the following steps: Retrieving and traversing a model database, wherein the model database includes multiple types of fruit varieties and a prediction model corresponding to each type of the fruit variety; the prediction model characterizes the relationship between the fruit skin hardness and the embryo development rate; If the model database contains the fruit variety corresponding to the target variety, the fruit skin hardness corresponding to the maximum embryo development rate is obtained through the prediction model; The peel hardness corresponding to the maximum embryo development rate is taken as the middle value of the target hardness range corresponding to the target variety, thereby obtaining the target hardness range.

[0033] For example, please refer to Figure 2 As shown in the figure, the prediction model is explained by taking the three target varieties of 'Flame Seedless', 'Ruby Seedless' and 'Jing Zao Jing' as examples. The fruit skin hardness of 'Flame Seedless' is taken as the x-axis and the embryo development rate is taken as the y-axis, and a quadratic regression equation is successfully constructed: y=4.60E-06x 2 +0.024x-4.921, R = 0.728. Pearson correlation coefficient analysis showed that the fruit skin hardness of 'Flame Seedless' was significantly correlated with its embryo development rate (r = 0.853, P ≤0.05) ( Figure 2 -A). Further calculations showed that when the peel hardness was 720 g, the embryo development rate reached the maximum, so the middle value of the target hardness range of 'Flame Seedless' was 720 g, and the corresponding time was the optimal sampling time for 'Flame Seedless' embryo-rescued fruits.

[0034] The peel hardness of 'Ruby Seedless' was taken as the x-axis and the embryo development rate as the y-axis, and a quadratic regression equation was successfully constructed: y=6.64E-06x 2 +0.014x-6.003, R = 0.838. Pearson correlation coefficient analysis showed that the fruit skin hardness of 'Ruby Seedless' was significantly correlated with its embryo development rate (r = 0.915, P ≤0.05) ( Figure 2 -B). It was further calculated that when the fruit skin hardness was 990 g, the embryo development rate reached the maximum, which was the best sampling time for the embryo-rescued fruit of 'Ruby Seedless'.

[0035] Taking the fruit skin hardness of 'Jing Zao Jing' as the x-axis and the embryo development rate as the y-axis, a quadratic regression equation was successfully constructed: y=-0.0002x 2 +0.272x-71.719, R = 0.953. Pearson correlation coefficient analysis showed that the fruit skin hardness of 'Jing Zao Jing' was significantly correlated with its embryo development rate (r = 0.811, P ≤0.05) ( Figure 2 -C). It was further calculated that when the fruit skin hardness was 633g, the embryo development rate reached the maximum, which was the best sampling time for embryo-rescued fruits of 'Jingzaojing'.

[0036] For 'flame without core', partial least squares regression (abbreviated as PLSR) in the training set ( R 2 =0.98, RMSE=22.73), Ridge Regression (abbreviation: RR) ( R 2 =0.98, RMSE=22.42), multiple linear regression (abbreviation: MLR) ( R 2 =0.98, RMSE=22.42), Random Forest Regression (abbreviation: RFR) ( R 2 =0.98, RMSE=22.22), Extreme Gradient Boosting (abbreviated as: XGBoost) ( R 2 =1, RMSE=0.03) has a better modeling effect, while the PLSR in the test set ( R 2 =0.94, RMSE=42.26), RR ( R 2 =0.94, RMSE=43.33) and MLR ( R 2 =0.94, RMSE=43.33) has good recognition effect, among which PLSR performs best. Therefore, it is generally believed that the target recognition model for "flame without core" is D1+PLSR. For "ruby without core", RR and MLR have the best recognition effect on the test set. R 2 is 0.79, and RMSE is 66.31. Considering that the MLR model calculation is simpler than the RR model, D1+MLR ( R 2 =0.79, RMSE=66.31) model as the target recognition model for 'Ruby Seedless'. The target recognition model for 'Jingzao Crystal' is similar to that for 'Flame Seedless', which is D1+PLSR. R 2 reached 0.93, and the RMSE was 47.92. For all seedless grape samples, MMN+PLSR ( R 2 =0.85, RMSE=87.21) as the universal recognition model has the best recognition effect. The best model test set of the remaining data R 2 All reached above 0.8, indicating that the target recognition model selected in this study has a good overall recognition effect.

[0037] In summary, when the peel hardness of 'Flame Seedless', 'Ruby Seedless', and 'Beijing Early Crystal' is 720 g, 990 g, and 633 g, respectively, the ovules are in the best state, and it is easier to obtain more developing embryos, which can be used as an important reference indicator for the optimal sampling time of embryo-rescued fruits. In order to eliminate the possible errors in the experiment, we allow the peel hardness value corresponding to the optimal sampling time to have a range of 20 g, that is, the peel hardness of 'Flame Seedless', 'Ruby Seedless', and 'Beijing Early Crystal' is 720±20 g, 990±20 g, and 633±20 g, respectively, which is the optimal sampling time for the corresponding varieties. Above this range, the ovules have not fully developed; below this range, the ovules begin to abort, and it is not easy to obtain developing embryos.

[0038] In a preferred embodiment, after retrieving and traversing the model database, the following steps are also included: If the model database does not contain the fruit variety corresponding to the target variety, the corresponding target hardness range is obtained through a universal recognition model.

[0039] In a preferred embodiment, if the model database does not contain the fruit variety corresponding to the target variety, after obtaining the corresponding target hardness range through the universal recognition model, the following steps are also included: According to the middle value of the target hardness range obtained by the general recognition model and the corresponding embryo development rate, an initial model is trained to obtain a specific prediction model; The specific prediction model corresponding to the target variety is stored in the model database.

[0040] Specifically, the intermediate values ​​and embryo development rate data given by the general recognition model are used as training samples, and the model parameters are adjusted to gradually adapt the model to the characteristics of the target variety. For example, in the neural network model, the weights between neurons are continuously adjusted through the back propagation algorithm to minimize the error between the predicted value and the actual value. Once the training is completed, a specific prediction model is obtained, which is stored in the model database in correspondence with the target variety so that it can be directly called the next time the variety is encountered.

[0041] Specifically, when there is no target variety in the model database, the use of the universal recognition model can provide a preliminary sampling guide for unknown varieties of fruit, avoiding the situation where the target hardness range cannot be determined due to the lack of a specific variety model. This greatly improves the applicability of the entire method and can cope with a variety of different fruit varieties. At the same time, by continuously training specific prediction models and storing them in the database, as the number of uses increases, the types of models in the database will become more and more abundant, and the flexibility of the method will continue to increase.

[0042] In a preferred embodiment, after taking the peel hardness corresponding to the maximum embryo development rate as the middle value of the target hardness range corresponding to the target variety, the method further includes the following steps: According to the target variety, a first preset range is obtained; The step of obtaining the target hardness range specifically comprises the following steps: The target hardness interval is taken as the target hardness interval.

[0043] Exemplarily, once the first preset range is determined, the middle value of the target hardness interval within the range is taken as the final target hardness interval. For example, if the middle value is H and the first preset range is ±2%, the target hardness interval is [0.98H, 1.02H].

[0044] Furthermore, obtaining a first preset range according to the target variety specifically includes the following steps: Obtaining a target hardness change rate corresponding to the target variety; The first preset range is calculated according to the target hardness change rate.

[0045] For example, assume that the target hardness change rate is R (the unit can be hardness unit / day), and there are T days left before the estimated sampling time. Assuming that the hardness change is approximately linear, the maximum value of the hardness increase within T days is R*T, and the maximum value of the decrease is also R*T (taking into account the possibility of external factors causing the hardness to decrease). Suppose the current peel hardness is H0, and the peel hardness corresponding to the maximum embryo development rate is H. A ratio value (R*T) / (H-H0) is obtained by calculation. If this ratio value is ±2%, then the first preset range can be set to ±2%. This method takes into account the current development status of the fruit and future change trends, so that the first preset range is more in line with the actual situation.

[0046] Furthermore, the obtaining of the target hardness change rate corresponding to the target variety specifically includes the following steps: Retrieving multiple real-time peel hardnesses of the target variety and the corresponding acquisition times from the historical database to obtain a reference database; the multiple real-time peel hardnesses in the reference database cover multiple growth stages of the fruit to be sampled; Specifically, the fruit skin hardness data of this variety measured at different time points are extracted from the historical database, including data from various stages such as the fruit setting period, the fruit expansion period, and the fruit beginning to change color period.

[0047] Based on the reference database, obtaining hardness change rates in multiple stages; Specifically, the hardness change rate of the corresponding stage can be obtained according to the growth stage, or the hardness change rate of each stage can be calculated according to the time interval and the difference in peel hardness at adjacent time points. Assuming that in a certain stage, the time is from t1 to t2, and the corresponding peel hardness changes from H1 to H2, then the hardness change rate of this stage is (H2 - H1) / (t2 - t1).

[0048] The target hardness change rate is obtained by weighted averaging the hardness change rates of each stage, wherein the harderness change rate of the stage closer to the current growth stage is given a higher weight.

[0049] Specifically, by considering the hardness changes of the target variety at different growth stages, the calculated target hardness change rate more accurately reflects the change trend of the hardness of the fruit of the variety. Calculating the first preset range based on this change rate can make the range more consistent with the actual development of the fruit, and improve the accuracy of determining the target hardness interval. For varieties with faster growth rates, the target hardness change rate is larger, and the corresponding first preset range will be wider, which can better adapt to its rapid change characteristics. In addition, the hardness change rates at different stages are weighted and averaged, and the change rates close to the current growth stage are given higher weights, which fully considers the dynamic nature of fruit growth. As the fruit develops, the most recent growth stage has a greater impact on the current state. This method can reflect the latest change trend of the fruit more promptly, so that the determined first preset range is more in line with the actual situation. If the fruit has abnormal hardness changes in the recent growth stage, this method can adjust the first preset range more quickly to adapt to such changes.

[0050] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and its core ideas of this application. The above is only the preferred implementation method of this application. It should be pointed out that due to the limitations of textual expression and the objective existence of infinite specific structures, ordinary technicians in this technical field can make several improvements, modifications or changes without departing from the principles of the present invention, and can also combine the above technical features in an appropriate manner; these improvements, modifications, changes or combinations, or the direct application of the inventive concept and technical solution to other occasions without improvement, should be regarded as the scope of protection of this application.

Claims

1. A method for determining the optimal sampling time for seedless grape embryo rescue fruit, characterized in that: The following steps are involved: Determine the target variety of the fruit to be sampled, and obtain the corresponding target hardness range according to the target variety; At intervals of a first preset time length, a near-infrared spectral reflectance curve of the fruit to be sampled is obtained in real time, and the real-time peel hardness of the fruit to be sampled is obtained according to the near-infrared spectral reflectance curve; If the real-time fruit skin hardness is within the target hardness range, the fruit to be sampled is sampled.

2. The method for determining the optimal sampling time for seedless grape embryo rescue fruit according to claim 1, characterized in that: After obtaining the real-time peel hardness of the fruit to be sampled according to the near-infrared spectral reflectance curve, the following steps are also included: The real-time peel hardness is stored in a historical data set corresponding to the moment before the current moment to obtain the historical data set corresponding to the current moment, wherein the historical data set includes several groups of the real-time peel hardness and the acquisition time corresponding to each group of the real-time peel hardness.

3. The method for determining the optimal sampling time for seedless grape embryo-saving fruit according to claim 2, characterized in that: After obtaining the real-time peel hardness of the fruit to be sampled according to the near-infrared spectral reflectance curve, the following steps are also included: If the real-time peel hardness does not reach the target hardness interval, determining whether the real-time peel hardness reaches a first preset threshold, the first preset threshold being less than a lower limit of the target hardness interval; If the real-time peel hardness reaches the first preset threshold, the development rate is retrieved and obtained based on the historical data set; According to the growth rate, a predicted sampling time is obtained, and then a first interval length between the current time and the predicted sampling time is obtained; If the first interval duration is less than or equal to the second preset duration, the first preset duration is shortened according to the first interval duration.

4. The method for determining the optimal sampling time for seedless grape embryo rescue fruit according to claim 1, characterized in that: The method of obtaining the real-time peel hardness of the fruit to be sampled according to the near-infrared spectral reflectance curve specifically comprises the following steps: According to the target variety, a target recognition model corresponding to the target variety is obtained; The near-infrared spectral reflectance curve is input into the target recognition model to obtain the real-time peel hardness.

5. The method for determining the optimal sampling time for seedless grape embryo-saving fruit according to claim 2, characterized in that: The step of obtaining the corresponding target hardness range according to the target variety specifically includes the following steps: Retrieving and traversing a model database, wherein the model database includes multiple types of fruit varieties and a prediction model corresponding to each type of the fruit variety; the prediction model characterizes the relationship between the fruit skin hardness and the embryo development rate; If the model database contains the fruit variety corresponding to the target variety, the fruit skin hardness corresponding to the maximum embryo development rate is obtained through the prediction model; The peel hardness corresponding to the maximum embryo development rate is taken as the middle value of the target hardness range corresponding to the target variety, thereby obtaining the target hardness range.

6. The method for determining the optimal sampling time for seedless grape embryo rescue fruit according to claim 5, characterized in that: After retrieving and traversing the model database, the following steps are also included: If the model database does not contain the fruit variety corresponding to the target variety, the corresponding target hardness range is obtained through a universal recognition model.

7. The method for determining the optimal sampling time for seedless grape embryo rescue fruit according to claim 6, characterized in that: If the model database does not contain the fruit variety corresponding to the target variety, after obtaining the corresponding target hardness range through the universal recognition model, the method further includes the following steps: According to the middle value of the target hardness range obtained by the general recognition model and the corresponding embryo development rate, an initial model is trained to obtain a specific prediction model; The specific prediction model corresponding to the target variety is stored in the model database.

8. The method for determining the optimal sampling time for seedless grape embryo-rescued fruit according to claim 5, characterized in that: After taking the peel hardness corresponding to the maximum embryo development rate as the middle value of the target hardness range corresponding to the target variety, the method further includes the following steps: According to the target variety, a first preset range is obtained; The step of obtaining the target hardness range specifically comprises the following steps: The target hardness interval is taken as the target hardness interval.

9. The method for determining the optimal sampling time for seedless grape embryo-rescued fruit according to claim 8, characterized in that: The step of obtaining a first preset range according to the target variety specifically includes the following steps: Obtaining the target hardness change rate corresponding to the target variety; The first preset range is calculated according to the target hardness change rate.

10. The method for determining the optimal sampling time for seedless grape embryo rescue fruit according to claim 9, characterized in that: The step of obtaining the target hardness change rate corresponding to the target variety specifically includes the following steps: Retrieving multiple real-time peel hardnesses of the target variety and the corresponding acquisition times from the historical database to obtain a reference database; the multiple real-time peel hardnesses in the reference database cover multiple growth stages of the fruit to be sampled; Based on the reference database, obtaining hardness change rates in multiple stages; The target hardness change rate is obtained by weighted averaging the hardness change rates of each stage, wherein the harderness change rate of a stage closer to the current growth stage is given a higher weight.

Images