Soilless culture vegetable yield prediction method and learning training platform
By establishing a vegetable yield model and a virtual platform, the problem of inaccurate yield prediction in the virtual planting platform was solved, accurate prediction and learning training of vegetable yield were achieved in a virtual environment, and the learners' practical ability was enhanced.
Patent Information
- Application Number
- CN202510788195.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-06-13
- Publication Date
- 2025-09-23
AI Technical Summary
In existing virtual planting platforms, planting measures cannot be linked to planting results, resulting in inaccurate vegetable yield prediction results.
A theoretical yield model for a single vegetable plant, a theoretical yield model per unit area, and a vegetable damage model were established. Combined with the influence of production conditions, the actual yield was calculated using the loss coefficient. A virtual plant factory platform was created using three-dimensional modeling technology to provide a learning and training environment.
It achieved accurate prediction of vegetable yield in a virtual environment, enhanced learners' practical ability and application of theoretical knowledge, and improved the flexibility and accuracy of the planting process.
Smart Images
Figure CN120688245A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of agricultural production simulation technology, and in particular to a soilless vegetable yield prediction method and a learning and training platform. Background Art
[0002] In recent years, my country has been vigorously developing smart agriculture and promoting the digital transformation of agriculture. Plant factories are currently the most representative application of high-level smart agriculture. Plant factories enable indoor soilless vegetable cultivation in vertical spaces, independent of natural conditions such as sunlight, soil, and climate. Instead, they rely on a variety of sensors and IoT devices to achieve fully automated vegetable production.
[0003] Many universities and research institutions are currently conducting vegetable cultivation research using plant factories. However, these facilities are limited, and the vegetable cultivation cycle is long, making them unable to meet the practical needs of most learners. Therefore, using a virtual simulation platform for virtual vegetable cultivation is a promising solution. However, current virtual cultivation platforms generally lack the ability to correlate planting measures with results, leading to inaccurate yield predictions after modifying planting measures during the simulation. Summary of the Invention
[0004] The embodiments of the present application provide a soilless vegetable yield prediction method and a learning and training platform to solve the problem of inaccurate vegetable yield prediction results in the prior art.
[0005] The present invention provides a method for predicting the yield of soilless vegetable cultivation, comprising:
[0006] Establish a theoretical yield model for a single vegetable plant under theoretical conditions;
[0007] Combined with the planting density of vegetables, a theoretical yield model per unit area is established based on the theoretical yield model of single plant;
[0008] A vegetable damage model is established based on the impact of production conditions on the degree of vegetable damage. The vegetable damage model includes a loss coefficient, which changes with production conditions.
[0009] The vegetable damage model is combined with the theoretical yield model per unit area to obtain the actual yield model;
[0010] Obtain the current production conditions, determine the current loss coefficient based on the current production conditions, bring the current loss coefficient into the actual yield model, and obtain the vegetable yield prediction result.
[0011] In one possible implementation, when the current production conditions change multiple times at different times, the current loss coefficient corresponding to the current production conditions after each change is determined in sequence, and the highest target yield of the previous stage is calculated in chronological order for each change of the current production conditions. The highest target yield of the previous stage is used as the initial yield of the current stage, and the loss coefficient of the current stage is substituted into the actual yield model to obtain the vegetable yield forecast result of the current stage.
[0012] In one possible implementation, the theoretical yield per unit area model is expressed as:
[0013]
[0014] Among them, P(t) is the yield per unit area at time t, P0 is the maximum yield per unit area, e is a natural constant, r is the growth rate, which is 0.21, and t0 is the time when the yield per unit area reaches half of the maximum yield per unit area.
[0015] In one possible implementation, the vegetable damage model is expressed as:
[0016] R(t)=k×t 2
[0017] Among them, R(t) is the degree of damage to vegetables at time t, and k is the loss coefficient.
[0018] In one possible implementation, the loss coefficient is expressed as:
[0019]
[0020] Among them, k0 is the maximum loss coefficient caused by the extreme value of the current production conditions, s is the parameter of the current production conditions, s0 and s3 are the minimum and maximum values of the production conditions allowed during the vegetable planting process, and s1 and s2 are the minimum and maximum values of the suitable parameters during the vegetable planting process.
[0021] In one possible implementation, when a single production condition becomes abnormal and the corresponding loss coefficient is less than or equal to the coefficient threshold, the least harmful abnormal growth state among multiple abnormal growth states corresponding to the production condition is obtained, and a corresponding abnormal growth state prediction result is generated; when a single production condition becomes abnormal and the corresponding loss coefficient is greater than the coefficient threshold, all abnormal growth states corresponding to the production condition are obtained, and a corresponding abnormal growth state prediction result is generated.
[0022] In one possible implementation, when multiple production conditions are abnormal at the same time, each production condition is judged in turn and the corresponding abnormal growth state is obtained, and the abnormal growth states corresponding to all production conditions are merged to generate the corresponding abnormal growth state prediction result.
[0023] The present application also provides a soilless cultivation learning and training platform, which includes:
[0024] The training task publishing module is used to publish the selected training tasks to learners;
[0025] The parameter acquisition module is used to obtain the parameters of the growth conditions input by the learner after viewing the training task;
[0026] Growth simulation module, used to simulate various stages of vegetable growth according to parameters;
[0027] Yield prediction module, used to predict the actual yield of vegetables using the above method;
[0028] The evaluation module is used to generate the learner's training results based on the difference between the actual output and the target output.
[0029] In one possible implementation, the growth simulation module uses a three-dimensional model to display the status of each device in the plant factory and the vegetables at each stage.
[0030] The soilless vegetable yield prediction method and learning and training platform in this application have the following advantages:
[0031] Using the vegetable yield prediction model, learners are asked to enter a simulated environment to set up the various environmental parameters and fertilizer supply of the plant factory, and continuously adjust them during the production process to ensure that the vegetables are in the required production conditions from planting to harvesting. This will obtain relatively accurate vegetable yield prediction results under the influence of production conditions, thereby achieving the purpose of examining the learners' mastery of theoretical knowledge. BRIEF DESCRIPTION OF THE DRAWINGS
[0032] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.
[0033] Figure 1 This is a flowchart of a method for predicting vegetable yield in soilless culture provided in an embodiment of the present application.
[0034] Figure 2 This is the single plant growth curve provided in the examples of this application.
[0035] Figure 3 Schematic diagram of the planting positions of lettuce on a 54-well planting plate provided in an embodiment of the present application.
[0036] Figure 4 This is the yield curve of lettuce per unit area under theoretical conditions provided in the embodiments of this application.
[0037] Figure 5 This is a curve showing the relationship between abnormal production conditions and the degree of lettuce damage provided in the examples of this application.
[0038] Figure 6 The relationship curve between the single production condition and the loss coefficient provided in the embodiment of this application.
[0039] Figure 7 This is the lettuce yield curve per unit area under the influence of production conditions provided in the examples of this application. DETAILED DESCRIPTION
[0040] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.
[0041] Figure 1 This is a flow chart of a method for predicting the yield of soilless vegetable culture provided in an embodiment of the present application. This embodiment of the present application provides a method for predicting the yield of soilless vegetable culture, including:
[0042] S100, establish a theoretical yield model for a single vegetable plant under theoretical conditions.
[0043] For example, lettuce (hereinafter referred to as romaine lettuce) has a typical growth cycle of 40-45 days, and the harvested fresh weight is 150-250g. Generally, lettuce goes through the following process from planting to harvesting:
[0044] Germination period: Soak the seeds in water for germination. It takes about 1.5-2 days for the seeds to turn white.
[0045] Seedling stage: sow white seeds in seedling trays, and it takes about 15 days for 3-5 true leaves to grow. It grows rapidly but has strict requirements on environmental conditions.
[0046] Vegetative growth period: Planting the seedlings in the cultivation module for growth is a critical period for nutrient accumulation, which takes about 25-30 days. Lettuce can be picked when it is nearing the end of the vegetative growth period.
[0047] Reproductive growth period: includes flower bud differentiation period, flowering period and fruiting period. Factory cultivation of lettuce will not enter this period.
[0048] Since the germination period has low environmental requirements, only successfully germinated seeds will be used for sowing, which has basically no direct impact on the yield results of the plant factory. Therefore, the germination period can be ignored in the lettuce growth law model of the plant factory.
[0049] According to the laws of plant physiology, the growth process of lettuce groups after sowing conforms to the law of logistic growth model. Under theoretical conditions (the plant factory is strictly automated and controlled, and the abnormal conditions of individual plants and the influence of the microenvironment are not considered), the group growth process is consistent with the growth process of a single plant. Therefore, the growth curve of a single lettuce plant is expressed as follows: Figure 2 shown.
[0050] S110, combining the planting density of vegetables and establishing a theoretical yield model per unit area based on the theoretical yield model of a single plant.
[0051] For example, according to field research, in a plant factory, lettuce production is first carried out on a seedling raising device. When the seedlings grow 3-5 true leaves after about 15 days of seedling raising, the seedling raising is completed, and high-quality seedlings are selected for planting in the cultivation module. After planting, the lettuce will grow to maturity and be picked and sold. This process generally takes 25-30 days. In other words, it is the lettuce after planting that determines the output of the plant factory, that is, the lettuce with a growth cycle of 15-45 days. Therefore, this application performs growth curve modeling and analysis on the lettuce growth process of this cycle.
[0052] Plant factories generally use planting modules for seedling planting production. Each module has several layers of planting plates. Each planting plate is a 54-hole plate with a size of 600mm*900mm. The planting standard is to plant one plant every other hole, that is, to plant 27 plants. By establishing a single plant growth model, the yield of the entire plant factory can be obtained by calculating the number of plants planted per unit area. The position of lettuce cultivation on the planting plate is as follows: Figure 3 As shown, the darker squares are holes where lettuce is planted.
[0053] The ideal yield per square meter of planted lettuce can be calculated using the area of the 54-well plate and the number of lettuces planted. The formula is as follows:
[0054] P0=S×27 / (h×w)×P
[0055] Among them, P0 represents the maximum yield of lettuce per unit area, in g; S represents the unit area, which is 1m in this calculation. 2 ; h represents the height of the planting plate, which is actually 0.6m; w represents the width of the planting plate, which is actually 0.9m; 27 represents the number of lettuces planted on a 54-hole planting plate; P represents the ideal fresh weight of a single lettuce plant in one planting, which is calculated as 200g in this calculation.
[0056] Substituting specific values into the above formula, we can get the following results:
[0057] P0=1m 2 ×27 / (0.6m×0.9m)×200g=10000g
[0058] That is, the ideal yield per unit area in a single planting cycle is 10,000 g.
[0059] Redraw the logistic growth curve of the lettuce population growth process, and the yield curve of lettuce per unit area is as follows: Figure 4 shown. Figure 4 The calculation formula of the logistic growth curve, that is, the theoretical yield per unit area model is expressed as:
[0060]
[0061] Where P(t) is the yield per unit area at time t; r is the growth rate, a constant. Based on the slope of the curve, we take r = 0.21; t0 is the midpoint of the curve, the time when the yield per unit area reaches half of the maximum yield per unit area. Again, based on the 45-day lettuce growth cycle, we take the value of 23; and e is a natural constant. Substituting time t (the number of days planted) into the above formula yields the ideal lettuce yield for that day.
[0062] S120, establishing a vegetable damage model based on the impact of production conditions on the degree of vegetable damage, wherein the vegetable damage model includes a loss coefficient, and the loss coefficient changes with production conditions.
[0063] For example, the ideal yield model per unit area calculates the ideal weight and ideal yield, which is the yield of lettuce when all production conditions are optimal. Next, we summarize and analyze the situations when production conditions are abnormal. We then develop a damage model for when lettuce growth is adversely affected and yield loss occurs. Substituting this into the ideal yield model formula, we determine the yield changes under abnormal production conditions.
[0064] For lettuce production companies, the growth status of lettuce has the following key points:
[0065] 1. The plants are healthy, with good growth status and growth rate, and no diseases or insect pests;
[0066] 2. The plant grows too slowly, which affects timely picking and increases cultivation costs;
[0067] 3. Malnutrition of plants, such as leaf loss, dwarfing, and weak leaves, results in fresh weight below the standard value or poor quality, leading to reduced yield;
[0068] 4. Excessive nutrition in the plants causes excessive growth, which results in lower fresh weight than the standard or poor quality, leading to reduced yield;
[0069] 5. The occurrence of pests and diseases leads to increased planting costs, partial loss (leaf drop), total loss (removal), etc., resulting in reduced production;
[0070] Vegetable production conditions in a plant factory encompass five main categories: environmental control, lighting control, nutrient solution preparation, water circulation, and production measures. Production measures are equipment that cannot be changed once the plant factory is built, and this model assumes they have no adverse impact on production. Therefore, only the first four factors are considered in determining the impact of the production environment on growth.
[0071] The impact of various production conditions in plant factories on lettuce growth is shown in the following table.
[0072] Table 1 Factors affecting plant factory production conditions
[0073]
[0074] Based on the above statistics on lettuce growth patterns and factors affecting lettuce conditions, lettuce may be adversely affected by improper production conditions during its growth process. A single improper production condition may lead to multiple abnormal conditions, such as slow growth, weak seedlings, and simultaneous disease. However, to simplify the establishment of the growth model, this application selects 1-2 of these factors for statistical analysis. A table of abnormal growth conditions is established as follows.
[0075] Table 2 Statistics of lettuce growth patterns
[0076]
[0077] When production conditions and measures are controlled at optimal conditions, there will be no impact on the growth curve of vegetables. When the conditions leave the optimal range, vegetable growth will begin to be affected, and individual vegetables will undergo various growth state changes. For the group, the direct impact is that the final yield will be affected.
[0078] There are also different degrees of abnormal production conditions, and the degree of output loss also varies accordingly. In actual production, this change is very complex, sometimes a straight line, sometimes an S curve, and it is impossible to fit it with a certain type of curve. In the virtual platform, this application simplifies it to a simulation close to reality to achieve the purpose of teaching and training. Therefore, the specific reduction ratio of damaged output can be represented by a parabola, such as Figure 5 shown. Figure 5 The curve in is expressed as follows:
[0079] R(t)=k×t2
[0080] Where R(t) is the degree of damage at time t, expressed in %, with a maximum value of 100, indicating total loss; k is the loss coefficient. The larger the value of k, the steeper the curve, indicating that the vegetables are damaged faster.
[0081] By substituting the above formula into the number of days, we can get the following result:
[0082] When k = 1, the total loss will reach 100% on the 10th abnormal day;
[0083] When k = 0.1, the total loss will reach 100% on the 30th day of the anomaly;
[0084] When k=0.05, the loss coefficient is small, indicating that some production conditions with minor errors were used at the beginning of planting, and the abnormality will reach 45% on the 30th day.
[0085] To simplify the model, this application assumes that the production loss pattern caused by abnormal production conditions is similar. The following takes ambient temperature as an example.
[0086] The suitable temperature for lettuce growth is 18-22℃. Temperatures above or below the suitable temperature will reduce the growth rate of the plant. When the ambient temperature is below 14℃, the growth slows down significantly. When it is below 4℃, it enters dormancy. When it is above 35℃, diseases and pests occur. When it is above 40 degrees, the leaves stop photosynthesis.
[0087] The loss coefficient curve associated with a single production condition is as follows: Figure 6 shown. Figure 6 The curve expression formula in is as follows:
[0088]
[0089] Among them, k0 is the maximum loss coefficient caused by the extreme value of the current production conditions; s is the parameter of the current production conditions; s0 and s3 are the minimum and maximum values of the production conditions allowed during the planting process, and the conditions exceeding the limit are no longer considered; s1 and s2 are the minimum and maximum values of the suitable parameters, and the conditions within the interval are all suitable conditions.
[0090] The suitable indicators for each production condition are different, and the impact of indicator changes on vegetable growth is also different, that is, the maximum values of the affected parameters are also different. Based on the previously summarized Table 2, this application has made the following production condition range and loss coefficient table.
[0091] Table 3 Production condition range and loss coefficient table
[0092] Production conditions <![CDATA[s0]]> <![CDATA[s1]]> <![CDATA[s2]]> <![CDATA[s3]]> <![CDATA[k0]]> 1 Ambient temperature (℃) 4 18 25 40 5 2 Ambient humidity (%) 0 60 80 100 1 3 <![CDATA[CO2 concentration (μmol.mol -1 )]]> 200 600 1000 1500 3 4 ventilation 0 1 2 4 1 5 <![CDATA[Light intensity (μmol.m -2 .s -1 )]]> 0 200 400 1000 5 6 Spectrum (ratio of red and blue light) 4 8 12 16 2 7 Photoperiod (h) 4 12 16 24 2 8 Nutrient solution configuration: EC value 0 2 2.5 4.5 5 9 pH 4 6 6.5 8 2 10 Intermittent cycle: pump on time (min) 0 10 20 120 0.5 11 Intermittent cycle: pump off time (min) 0 30 50 120 2 12 Liquid temperature(℃) 4 18 22 40 3 13 <![CDATA[Oxygen content (mg / L -1 )]]> 0 4 5 8 2
[0093] The following rules can be calculated from the above table:
[0094] When production conditions deviate from the suitable zone by about 10%, the loss coefficient is 0.01-0.05. Substituting this into the damage degree calculation formula, the loss ratio on the 30th day of abnormality is 9-45%;
[0095] When the production conditions deviate from the suitable zone by about 20%, the loss coefficient is 0.02-0.1. Substituting it into the damage degree calculation formula, the loss ratio on the 30th day of abnormality is 18-90%.
[0096] S130, combining the vegetable damage model with the theoretical yield per unit area model to obtain an actual yield model.
[0097] For example, after obtaining the boundary values and loss coefficients of various production conditions, the loss coefficients can be incorporated into the vegetable damage model to obtain a complete vegetable damage model. By substituting this damage model into the theoretical yield model per unit area, the final actual yield can be evaluated.
[0098] Substitute k into the damage degree formula:
[0099] R(t)=k×(t-t1) 2
[0100] Among them, t1 represents the starting time of production loss, that is, the date when the production conditions become abnormal.
[0101] Then the theoretical output multiplied by the percentage of damage is the actual output:
[0102]
[0103] Among them, P1(t) represents the unit area output at time t under the influence of current production conditions.
[0104] Substitute the formula of P(t) into the above model:
[0105]
[0106] The above formula is the actual output model.
[0107] Assuming t1 = 20, k = 0.05, the actual yield per unit area curve after fitting is as follows: Figure 7 shown.
[0108] The above calculation assumes that the learner modified a production condition on a particular day, generating a loss coefficient. In addition to modifying the parameters of a single production condition, the learner may also modify the parameters of multiple production conditions simultaneously. Observing the relationship between P1(t) and k in the actual lettuce yield model, we can see that the two are a linear function, indicating that multiple loss coefficients can be added together. The new loss coefficient after addition can be expressed as:
[0109] k=k1+k2+……+k 13
[0110] Among them, k1, k2...k 13 It represents the loss coefficients corresponding to multiple different production conditions, and the overall loss coefficient k is obtained by adding them together.
[0111] The above calculations assume that the learner only modified the production condition parameters once. In practice, learners may modify production conditions multiple times at different time points. To address this situation, we sequentially determine the current loss coefficient corresponding to each change in the current production condition. We then calculate the highest target yield for each change in the current production condition, using the previous target yield as the initial yield for the current stage. We then apply the current stage's loss coefficient to the actual yield model to obtain the vegetable yield forecast for the current stage.
[0112] Specifically, first, a table needs to be created to record the number of days each production condition change occurs and the resulting loss coefficient. Learners can calculate k using the formula for the loss coefficient k previously described. The table is sorted chronologically, allowing for the continuous insertion of new adjustment records.
[0113] Table 4 Time-loss coefficient record table
[0114]
[0115]
[0116] As can be seen from the above table, the learner modified the production conditions to the unsuitable range on the 15th and 18th days, which led to an anomaly and generated a loss coefficient. On the 25th day, the production conditions were modified to the suitable range, and the loss coefficient returned to 0. On the 32nd day, the production conditions were modified again to abnormal conditions, and a new loss coefficient was generated.
[0117] When there are losses in the production process, the target yield of lettuce has been decreasing. When a new loss coefficient is generated, the target maximum yield needs to be recalculated based on the current loss level.
[0118] Assuming that we are currently at the first time point, we can use the actual production model to calculate the target production at the first time point.
[0119] Assuming we are currently at the second time point, to draw the stage yield curve, we must first calculate the maximum target yield for the second stage. This yield is calculated by multiplying the loss level at the end of the first stage by the maximum target yield for the first stage. The formula is as follows:
[0120]
[0121] Among them, k1 represents the previous time point, or the loss coefficient of the previous stage, which can be obtained from the record table; t2 represents the start time of the second stage, that is, the end time of the previous stage, which can also be obtained from the record table.
[0122] Based on this target output P1, the output curve of the current second stage can be calculated:
[0123]
[0124] After substituting the calculation formula of P1:
[0125]
[0126] The method for calculating multiple changes in the loss coefficient is to iterate over all time points and perform multiple calculations to ultimately derive the actual yield model curve. Based on the above calculation rules, the target yield can be repeatedly calculated and P0, P1, P2, etc. in the formula can be updated, while the rest of the formula remains unchanged.
[0127]
[0128] When all production conditions enter the appropriate range, k n =0, does not affect the use of the above formula.
[0129] At this point, the actual yield model for lettuce production in a plant factory has been established.
[0130] S140, obtaining current production conditions, determining a current loss coefficient based on the current production conditions, and bringing the current loss coefficient into the actual yield model to obtain a vegetable yield prediction result.
[0131] For example, when production conditions are abnormal, the growth state is abnormal, resulting in reduced yield. The present application also queries and obtains the growth state prediction of vegetables through Table 2.
[0132] For example, when abnormal production conditions include low ambient humidity, high temperature, and high pH of the nutrient solution, the abnormal conditions caused by these conditions can be found as follows.
[0133] Table 5 Abnormal growth status
[0134]
[0135] These abnormal growth states will not all occur at the same time, but 1-2 with the highest probability of occurrence will be selected as the abnormal growth state prediction results based on the degree of abnormality.
[0136] The rules are as follows:
[0137] 1. When a single production condition is abnormal and the corresponding loss coefficient is less than or equal to the coefficient threshold, the least harmful abnormal growth state among multiple abnormal growth states corresponding to the production condition is obtained, and the corresponding abnormal growth state prediction result is generated. For example, if the ambient temperature is 27°C and the loss coefficient is 0.08, only slow growth is predicted, and weak seedlings are not predicted.
[0138] 2. When a single production condition is abnormal and the corresponding loss coefficient is greater than the coefficient threshold, all abnormal growth states corresponding to the production condition are obtained and corresponding abnormal growth state prediction results are generated. For example, if the ambient humidity is less than 40%, slow growth and dry heart disease can be predicted simultaneously.
[0139] 3. If multiple production conditions are abnormal at the same time, each production condition is judged in turn and the corresponding abnormal growth status is obtained. The abnormal growth status corresponding to all production conditions is merged to generate the corresponding abnormal growth status prediction result.
[0140] Based on the above-mentioned vegetable yield prediction method, this application also provides a soilless cultivation learning and training platform. Using 3D modeling technology, a virtual plant factory and 3D models of key equipment related to vegetable cultivation are created, along with 3D models of each stage of vegetable cultivation, from seeding to harvest. This platform allows learners to perform related planting tasks.
[0141] This platform simulates the plant factory's environmental control system, lighting system, nutrient solution system, irrigation system, etc. Through this platform, learners can learn the entire process of vegetable cultivation in a plant factory, realize interactive operation and parameter control, and enhance their learning experience and practical skills.
[0142] The core functions of this platform are as follows:
[0143] 1. In a virtual planting environment, the entire process of a certain vegetable from sowing to harvesting is realized, and the changes in the appearance of the plant at each major stage from germination to harvest can be observed;
[0144] 2. Develop a data monitoring platform to collect and display sensor data such as temperature and humidity, light intensity, and nutrient solution EC value within the plant factory, providing learners with accurate planting decision support;
[0145] 3. Provide an automated control platform, which can be used to adjust the parameters of all environmental control, lighting, nutrient solution circulation and other equipment in the production environment;
[0146] 4. Design planting tasks, providing planting tasks by date and stage, such as germination, sowing, planting, environmental control, pest and disease control, etc. You can also set tasks for abnormal growth conditions at any time and require learners to correct them. After entering the task, learners need to monitor the production environment, allocate nutrient solution, and configure various environmental and nutrient solution circulation parameters to ensure that vegetables can be produced under the most suitable conditions and complete this task;
[0147] 5. Based on the actual vegetable yield model established above, the function of calculating expected yield is provided to realize the functions of vegetable yield prediction and planting task execution result judgment.
[0148] The platform provides a complete task operation mode. The task operation process is as follows:
[0149] After selecting a training task, learners first enter a 3D visual production environment to understand the vegetable growth status. They also access the data monitoring platform to observe various sensor data to understand the current production date and growth status of the vegetables. Based on this data, learners make decisions about vegetable production based on their knowledge. Based on the results of these decisions, they adjust the environment, lighting, and nutrient solution circulation to the appropriate level, and then prepare the appropriate nutrient solution and add it to the nutrient solution circulation system.
[0150] If the learner sets it improperly and the parameters of a certain production condition do not enter the appropriate range, the system will also execute the setting.
[0151] When all settings are completed, the system will judge these environment and nutrient solution settings, and calculate the loss coefficient k1-k corresponding to each production condition one by one according to the production condition range designed by the system's built-in production model and Table 3 and the loss coefficient calculation formula. n The combined loss coefficient k is obtained and recorded in Table 5.
[0152] After obtaining the necessary parameters for calculating yield changes, such as the target vegetable yield, the last production parameter modification date, the last loss coefficient, and the current production date from the data monitoring platform, the yield change calculation can be made based on the actual yield model for this production condition parameter modification. At the same time, the current yield change curve can be drawn and displayed on the data monitoring platform.
[0153] The system can also predict vegetable diseases and pests based on the production condition parameter settings, and the prediction results will also be recorded in the task execution results.
[0154] Based on whether the parameter adjustment results of this production condition are within the appropriate range and the final output changes, it can be judged whether the learner's task execution is qualified and the performance judgment result can be given.
[0155] A: All production parameters and nutrient solution configurations are within the appropriate range, and the target yield is consistent with the actual yield.
[0156] In B: Some production parameters deviate from the optimal range, but this has little impact on actual yield, and the final yield loss is predicted to be less than 10%;
[0157] C Poor: Some production parameters deviate from the appropriate range, which has a significant impact on actual production. The final production loss is expected to be 10-50%;
[0158] D Failed: The parameters of some production conditions deviate from the appropriate range, which has a significant impact on the actual output. The final output forecast loss exceeds 50%. In this case, the learner is required to perform the task again.
[0159] By combining the actual vegetable yield model with the virtual simulation platform, the rigid training mode of the single process of the virtual simulation training platform has been broken, and most of the settings of various production equipment and control equipment in the plant factory can be opened to learners for setting, greatly enhancing the openness and flexibility of the platform.
[0160] Since the impact of each production condition on the change in output is relatively subtle, computer floating-point operations often result in inaccurate calculation results due to insufficient precision. The method of the present application does not need to calculate the losses caused by each setting one by one, but can calculate them together, thus avoiding inaccurate calculation of output losses.
[0161] This comprehensive calculation method also makes learners more careful and patient when adjusting production parameters, so that all parameters can be adjusted into the appropriate range. It also forces learners to strengthen their theoretical knowledge and apply it to the virtual simulation training platform, thereby enhancing the effect of combining theory with practice.
[0162] Although the preferred embodiments of the present application have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present application.
[0163] Obviously, those skilled in the art may make various changes and modifications to this application without departing from the spirit and scope of this application. Thus, if these modifications and variations of this application fall within the scope of the claims of this application and their equivalents, this application is intended to include these modifications and variations.
Claims
1. A method for predicting the yield of soilless cultured vegetables, characterized in that: include: Establish a theoretical yield model for a single vegetable plant under theoretical conditions; In combination with the planting density of vegetables, a theoretical yield per unit area model is established according to the theoretical yield per plant model; Establishing a vegetable damage model based on the impact of production conditions on the degree of vegetable damage, wherein the vegetable damage model includes a loss coefficient, and the loss coefficient changes with the production conditions; Combining the vegetable damage model with the theoretical yield per unit area model to obtain an actual yield model; The current production conditions are obtained, a current loss coefficient is determined according to the current production conditions, and the current loss coefficient is brought into the actual yield model to obtain a vegetable yield prediction result.
2. The method for predicting yield of soilless vegetable culture according to claim 1, wherein: When the current production conditions change multiple times at different times, the current loss coefficient corresponding to the current production conditions after each change is determined in turn, and the highest target yield of the previous stage when the current production conditions change each time is calculated in chronological order. The highest target yield of the previous stage is taken as the initial yield of the current stage, and the loss coefficient of the current stage is substituted into the actual yield model to obtain the vegetable yield prediction result of the current stage.
3. The method for predicting yield of soilless vegetable culture according to claim 1, wherein: The theoretical yield per unit area model is expressed as: Among them, P(t) is the yield per unit area at time t, P0 is the maximum yield per unit area, e is a natural constant, r is the growth rate, which is 0.21, and t0 is the time when the yield per unit area reaches half of the maximum yield per unit area.
4. The method for predicting yield of soilless vegetable culture according to claim 1, wherein: The vegetable damage model is expressed as: R(t)=k×t 2 Among them, R(t) is the degree of damage to vegetables at time t, and k is the loss coefficient.
5. The method for predicting yield of soilless cultured vegetables according to claim 4, characterized in that: The expression of the loss coefficient is: Among them, k0 is the maximum loss coefficient caused by the extreme value of the current production conditions, s is the parameter of the current production conditions, s0 and s3 are the minimum and maximum values of the production conditions allowed during the vegetable planting process, and s1 and s2 are the minimum and maximum values of the suitable parameters during the vegetable planting process.
6. The method for predicting yield of soilless vegetable culture according to claim 1, wherein: When a single production condition is abnormal and the corresponding loss coefficient is less than or equal to a coefficient threshold, obtaining the least harmful abnormal growth state among multiple abnormal growth states corresponding to the production condition, and generating a corresponding abnormal growth state prediction result; When a single production condition is abnormal and the corresponding loss coefficient is greater than the coefficient threshold, all the abnormal growth states corresponding to the production condition are obtained and corresponding abnormal growth state prediction results are generated.
7. A method for predicting yield of soilless cultured vegetables according to claim 6, characterized in that: When multiple production conditions are abnormal at the same time, each production condition is judged in turn and the corresponding abnormal growth state is obtained, and the abnormal growth states corresponding to all the production conditions are merged to generate the corresponding abnormal growth state prediction result.
8. A soilless cultivation learning and training platform, characterized in that: include: The training task publishing module is used to publish the selected training tasks to learners; A parameter acquisition module, used to acquire the parameters of the growth conditions input by the learner after viewing the training task; A growth simulation module is used to simulate various stages of vegetable growth according to the parameters; A yield prediction module, configured to predict the actual yield of vegetables using the method according to any one of claims 1 to 7; The evaluation module is used to generate a training score of the learner according to the difference between the actual output and the target output.
9. The soilless cultivation learning and training platform according to claim 8, characterized in that: The growth simulation module uses a three-dimensional model to display the status of each device in the plant factory and the vegetables at each stage.