Potato soilless culture system and method based on multi-component nutrient solution control

Abstract

The embodiment of the invention discloses a potato soilless culture system and method based on multi-component nutrient solution control, and the system comprises a data monitoring module, a scheme regulation and control module and a scheme execution optimization module, comprising growth monitoring data, environment monitoring data and nutrient solution monitoring data; the scheme regulation and control module is used for obtaining nutrient solution demand data in combination with potato cultivation data according to a convolutional neural network-gated cycle unit (CNN-GRU) model, and generating a dynamic nutrient solution regulation and control strategy in combination with the nutrient solution demand data and nutrient solution monitoring data according to a deep reinforcement learning (DRL) model; and the scheme execution optimization module is used for executing a dynamic nutrient solution regulation strategy, regulating environment monitoring data and a nutrient solution formula, and optimizing the nutrient solution formula according to nutrient solution parameters and potato growth parameters. According to the application, the precision, intelligence and high efficiency of potato cultivation can be improved.

Description

Technical Field

[0001] The present application relates to the field of soilless cultivation technology based on machine learning, and is related to but not limited to a potato soilless cultivation system and method based on multi-component nutrient solution control. Background Art

[0002] As an important food and cash crop, potatoes boast numerous advantages, including strong adaptability, a short growth cycle, high yield, and rich nutrition. They play a crucial role in ensuring food security and promoting agricultural economic development. Soilless cultivation techniques for potato cultivation better meet the precise environmental and nutrient requirements of potato growth, effectively avoiding the soil-borne diseases and continuous cropping problems that commonly occur with traditional soil cultivation. While also improving potato yield and quality, they enable efficient, automated, and intelligent potato cultivation.

[0003] Precise regulation of the nutrient solution is a key factor in soilless cultivation. Existing technologies often use general or modified nutrient solution formulas (such as the Hoagland formula) to meet the needs of potatoes at different growth stages, adjusting the types and proportions of nutrients to meet their growth and development requirements. Nutrient solution management and control involves monitoring and adjusting parameters such as temperature, pH, and conductivity through manual monitoring or simple automatic control systems to maintain stability and suitability. Intelligent nutrient solution control uses empirical models or single machine learning algorithms to predict nutrient needs and optimize relevant parameters through big data analysis. Potatoes at different growth stages have different requirements for nutrient solution composition and proportions. However, existing technologies often suffer from poorly targeted nutrient solution formulations, limited precision in nutrient solution management and control, and limited intelligence. These technologies are unable to meet the dynamic needs of potatoes at different growth stages, which to some extent limits the further development and widespread application of soilless potato cultivation.

[0004] Therefore, there is an urgent need for a new potato soilless cultivation system and method based on multi-component nutrient solution control to solve the problems existing in existing potato soilless cultivation related technologies, such as weak nutrient solution formulation targeting, low control accuracy and insufficient intelligence level, so as to achieve precise dynamic control of the potato growth environment and nutrient solution, thereby improving the efficiency and actual benefits of potato soilless cultivation, and promoting the sustainable development of modern agriculture. Summary of the Invention

[0005] The embodiments of the present application provide a potato soilless cultivation system and method based on multi-component nutrient solution control.

[0006] The technical solution of the embodiment of the present application is implemented as follows: In a first aspect, an embodiment of the present application provides a potato soilless cultivation system based on multi-component nutrient solution control, the system comprising a data monitoring module, a scheme control module, and a scheme execution optimization module, wherein: The data monitoring module is used to monitor and collect potato cultivation data in real time, wherein the potato cultivation data includes growth monitoring data, environmental monitoring data, and nutrient solution monitoring data. The solution control module is used to obtain nutrient solution demand data based on the CNN-GRU model and the potato cultivation data, and to generate a dynamic nutrient solution control strategy based on the DRL model and the nutrient solution demand data and the nutrient solution monitoring data. The solution execution optimization module is used to execute the dynamic nutrient solution control strategy, control the environmental monitoring data and the nutrient solution formula, and optimize the nutrient solution formula based on nutrient solution parameters and potato growth parameters.

[0007] The technical solution provided by the present application monitors and collects potato cultivation data in real time through a data monitoring module. The potato cultivation data includes growth monitoring data, environmental monitoring data and nutrient solution monitoring data. The potato cultivation data of various aspects is accurately obtained through a variety of methods, and the factors affecting potato growth are considered more comprehensively, thereby improving the accuracy of the subsequent control scheme obtained based on the potato cultivation data; through the scheme control module, according to the CNN-GRU model, the nutrient solution demand data is obtained in combination with the potato cultivation data, so as to achieve accurate prediction of the nutrient solution demand of potatoes at different growth stages; according to the DRL model, the nutrient solution demand data and the nutrient solution monitoring data are combined to generate a dynamic nutrient solution control strategy, which is convenient for flexible adjustment according to the potato cultivation conditions; the dynamic nutrient solution control strategy is executed through the scheme execution optimization module The system uses a closed-loop control strategy to adjust environmental monitoring data and nutrient solution formula to make the nutrient solution supply and environmental conditions more in line with the actual growth needs of potatoes, avoid problems such as overnutrition, undernutrition and unsuitable environment, thereby promoting the healthy growth of potatoes and improving potato yield and quality. In addition, the nutrient solution formula is optimized according to the nutrient solution parameters and potato growth parameters. Through this full-cycle closed-loop iteration of "data collection-implementation plan-effect feedback-system optimization", the control strategy is accurately implemented, and it is continuously iterated to optimize the control strategy so that the control strategy can adapt to the different growth stages and growth environment of potatoes. Through this closed-loop feedback correction mechanism, the accuracy and sensitivity of nutrient solution parameter control can be improved, and labor costs can be greatly reduced, thereby realizing the precision, intelligence and efficiency of potato cultivation controlled by nutrient solution.

[0008] Optionally, the scheme regulation module comprises a nutrient solution prediction unit and a scheme regulation unit, wherein: the nutrient solution prediction unit is configured to analyze the growth monitoring data according to a CNN model, determine a potato growth stage, output a growth stage one-hot encoding vector, combine the environment monitoring data, the nutrient solution monitoring data and the growth stage one-hot encoding vector, and obtain the nutrient solution demand data according to a gated recurrent unit model; and the scheme regulation unit is configured to fuse the nutrient solution demand data and the nutrient solution monitoring data according to a DRL model, and generate the dynamic nutrient solution regulation strategy. Optionally, the step of analyzing the growth monitoring data according to the CNN model, determining a potato growth stage, and outputting a growth stage one-hot encoding vector comprises: preprocessing and normalizing the growth monitoring data to obtain processed growth monitoring data; inputting the processed growth monitoring data into the CNN model, scanning the processed growth monitoring data through a bottom convolution kernel, and identifying potato growth feature data; performing down-sampling on the potato growth feature data through a pooling layer, removing repeated feature information, and retaining core growth feature data; converting the core growth feature data into a probability distribution of each growth stage through a multi-scale fusion technology and a normalization exponential function, wherein the growth stages include a germination stage, a seedling stage, a tuber formation stage, a tuber bulking stage and a mature stage; taking a growth stage with the largest probability value as a current growth stage, and converting the probability distribution of the current growth stage into the growth stage one-hot encoding vector.

[0009] Optionally, the combination of the environmental monitoring data, the nutrient solution monitoring data and the growth stage one-hot encoding vector, and the nutrient solution demand data according to the gated cyclic unit model, includes: encoding each growth stage, the environmental monitoring data and the nutrient solution monitoring data into a five-dimensional growth stage vector, a four-dimensional environmental data vector and a twelve-dimensional nutrient solution data vector, splicing the five-dimensional growth stage vector, the four-dimensional environmental data vector and the twelve-dimensional nutrient solution data vector to obtain a twenty-one-dimensional feature vector; constructing the twenty-one-dimensional feature vector into a time series input sequence according to the daily time step, the time series input sequence including the number of samples, the time step and the feature dimension, the time step is 24, the The feature dimension is 21; the time series input sequence is input into the gated recurrent unit model, and in the update gate, the linear combination value of the current input sequence and the historical hidden state sequence is calculated using a logical function to dynamically retain historical valid information; in the reset gate, the weights of historical information to be ignored in the historical valid information and the response abnormal data are screened and calculated, the historical information is combined with the current input sequence, and the candidate hidden state sequence is generated by activation through the hyperbolic tangent activation function; in the update gate, the candidate hidden state sequence is weightedly fused with the historical hidden state sequence to obtain the current hidden state sequence; the hidden state sequence of the final time step is mapped through the fully connected layer to obtain the nutrient solution demand data of the current growth stage.

[0010] Optionally, the nutrient solution demand data and the nutrient solution monitoring data are integrated according to the DRL model to generate the dynamic nutrient solution control strategy, including: in the DRL model, a state space, an action space and a reward function are constructed according to the nutrient solution demand data, the nutrient solution monitoring data and the growth stage one-hot encoding vector, and the calculation formula of the reward function is expressed by the following formula: ; Where, represents the reward function; Indicates the difference between the actual adjustment amount and the target adjustment amount; Indicates the nutrient solution demand data; Indicates the nutrient concentration matching reward; Indicates the difference in pH before and after regulation; Indicates the difference between actual and target conductivity; represents the target conductivity during the growth phase; Indicates conductivity concentration stability bonus; Indicates the pump running time; 、 、 、 They respectively represent the weight coefficients of the nutrient concentration matching reward, the pH stability reward, the conductivity concentration stability reward and the pump running time; in the executor (Actor) network, the state space is received and the action space is output, the action space is executed and the real-time nutrient solution parameters are obtained, and the reward value is calculated in combination with the reward function; in the evaluator (Critic) network, the evaluation value between the state space and the action space is calculated, the value error is calculated according to the reward function and the temporal difference algorithm, and the critic network parameters are updated according to the value error; the gradient direction of the evaluation value is input to the actor network through the critic network, the actor network parameters are updated, and the dynamic nutrient solution control strategy is obtained.

[0011] Optionally, the scheme execution optimization module includes a multi-liquid storage tank collaborative control unit, an environmental auxiliary control unit and a feedback correction unit, wherein: the multi-liquid storage tank collaborative control unit is used to control the nutrient solution formula through a pulse width modulation metering pump according to the dynamic nutrient solution control strategy; the environmental auxiliary control unit is used to control the environmental monitoring data through a fill light device, a ventilation device and a carbon dioxide control device according to the dynamic nutrient solution control strategy; the feedback correction unit is used to calculate the potato growth rate according to the regulated potato growth parameters, calculate the nutrient solution parameter correction amount according to the regulated nutrient solution parameters and the proportional-integral-differential control algorithm, and optimize the nutrient solution formula according to the potato growth rate and the nutrient solution parameter correction amount.

[0012] Optionally, the growth monitoring data is obtained by regularly shooting with cameras placed above and to the sides of the potato cultivation area; the environmental monitoring data is obtained by real-time monitoring with temperature sensors, light intensity sensors and carbon dioxide sensors installed on the potato plants; the nutrient solution monitoring data is obtained by monitoring with high-precision sensors in the circulation pipeline, and the high-precision sensors include ion-selective electrobiosensors, pH sensors and conductivity sensors.

[0013] In a second aspect, the embodiments of the present application provide a potato soilless cultivation method based on multi-component nutrient solution control. The method is applied to a potato soilless cultivation system based on multi-component nutrient solution control. The system comprises a data monitoring module, a scheme control module and a scheme execution optimization module. The method comprises the following steps: real-time monitoring and collecting potato cultivation data, wherein the potato cultivation data comprises growth monitoring data, environmental monitoring data and nutrient solution monitoring data; obtaining nutrient solution demand data according to a CNN-GRU model in combination with the potato cultivation data, generating a dynamic nutrient solution control strategy according to a DRL model in combination with the nutrient solution demand data and the nutrient solution monitoring data; executing the dynamic nutrient solution control strategy to control the environmental monitoring data and the nutrient solution formula, and optimizing the nutrient solution formula according to nutrient solution parameters and potato growth parameters.

[0014] In a third aspect, the embodiments of the present application provide an electronic device comprising a memory and a processor. The memory stores a computer program capable of running on the processor. When the processor executes the program, the steps of the above-described potato soilless cultivation method based on multi-component nutrient solution control are implemented.

[0015] In a fourth aspect, the embodiments of the present application provide a computer readable storage medium having a computer program stored thereon. When the processor executes the program, the steps of the above-described potato soilless cultivation method based on multi-component nutrient solution control are implemented.

[0016] The technical solutions provided by the embodiments of the present application have at least the following beneficial effects: The present application provides a potato soilless cultivation system and method based on multi-component nutrient solution control. The data monitoring module monitors and collects potato cultivation data in real time. The potato cultivation data includes growth monitoring data, environmental monitoring data and nutrient solution monitoring data. The potato cultivation data of various aspects are accurately acquired through multiple methods, and the factors affecting potato growth are considered more comprehensively, thereby improving the accuracy of the subsequent control scheme obtained based on the potato cultivation data. The scheme control module obtains nutrient solution demand data based on the CNN-GRU model and the potato cultivation data, thereby accurately predicting the nutrient solution demand of potatoes at different growth stages. The DRL model is used to combine the nutrient solution demand data and the nutrient solution monitoring data to generate a dynamic nutrient solution control strategy, which is convenient for flexible adjustment according to the potato cultivation conditions. The scheme execution optimization model is used to optimize the potato cultivation data. The block executes dynamic nutrient solution control strategy, regulates environmental monitoring data and nutrient solution formula, so that the nutrient solution supply and environmental conditions are more in line with the actual growth needs of potatoes, avoiding problems such as overnutrition, malnutrition and unsuitable environment, thereby promoting the healthy growth of potatoes and improving potato yield and quality. In addition, the nutrient solution formula is optimized according to the nutrient solution parameters and potato growth parameters. Through this full-cycle closed-loop iteration of "data collection-execution plan-effect feedback-system optimization", the control strategy is accurately implemented, and it is continuously iterated to optimize the control strategy so that the control strategy can adapt to the different growth stages and growth environment of potatoes. Through this closed-loop feedback correction mechanism, the accuracy and sensitivity of nutrient solution parameter control can be improved, and labor costs are greatly reduced, thereby realizing the precision, intelligence and efficiency of potato cultivation controlled by nutrient solution. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present application. Those skilled in the art can also derive other drawings based on these drawings without inventive work, among which: Figure 1 A schematic diagram of a potato soilless cultivation system based on multi-component nutrient solution control provided in an embodiment of the present application; Figure 2 A detailed schematic diagram of a potato soilless cultivation system based on multi-component nutrient solution control provided in an embodiment of the present application; Figure 3 A flow chart of a potato soilless cultivation method based on multi-component nutrient solution control provided in an embodiment of the present application; Figure 4 A hardware entity diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0018] In order to make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all of the embodiments. The following examples are used to illustrate the present application, but are not intended to limit the scope of the present application. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0019] In the following description, reference is made to “some embodiments”, which describes a subset of all possible embodiments, but it will be understood that “some embodiments” may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict.

[0020] It should be pointed out that the terms "first\second\third" involved in the embodiments of the present application are merely used to distinguish similar objects and do not represent a specific ordering of the objects. It can be understood that "first\second\third" can be interchanged with a specific order or sequence where permitted, so that the embodiments of the present application described here can be implemented in an order other than that illustrated or described here.

[0021] Those skilled in the art will understand that, unless otherwise defined, all terms used herein (including technical and scientific terms) have the same meaning as generally understood by those skilled in the art in the art to which the embodiments of the present application belong. It should also be understood that terms such as those defined in common dictionaries should be understood to have meanings consistent with their meanings in the context of the prior art, and will not be interpreted in an idealized or overly formal sense unless specifically defined as herein.

[0022] The embodiments of the present application will be further described below with reference to the accompanying drawings.

[0023] In view of the current problems in the field of soilless cultivation technology for potato soilless cultivation based on multi-component nutrient solution control, the embodiments of the present application provide a potato soilless cultivation system and method based on multi-component nutrient solution control.

[0024] The technical solution of the present application is introduced below, and first the system embodiment of the present application is introduced.

[0025] Please refer to Figure 1 , which shows a schematic diagram of a potato soilless cultivation system based on multi-component nutrient solution control provided in an embodiment of the present application, such as Figure 1As shown, the system comprises a data monitoring module 01, a scheme regulation module 02 and a scheme execution optimization module 03. The scheme execution optimization module 03 is connected with the monitoring module 01 and the scheme regulation module 02 respectively, and the data monitoring module 01, the scheme regulation module 02 and the scheme execution optimization module 03 are connected in sequence. The data monitoring module 01 is used for real-time monitoring and collecting potato cultivation data, including growth monitoring data, environmental monitoring data and nutrient solution monitoring data; the scheme regulation module 02 is used for obtaining nutrient solution demand data according to the CNN-GRU model combined with the potato cultivation data, and generating a dynamic nutrient solution regulation strategy according to the DRL model combined with the nutrient solution demand data and the nutrient solution monitoring data; the scheme execution optimization module 03 is used for executing the dynamic nutrient solution regulation strategy, regulating the environmental monitoring data and the nutrient solution formula, and optimizing the nutrient solution formula according to the nutrient solution parameters and the potato growth parameters.

[0026] In the embodiment of the present application, the data monitoring module 01 is used for real-time monitoring and collecting potato cultivation data, including growth monitoring data, environmental monitoring data and nutrient solution monitoring data. Specifically, the growth monitoring data is obtained by periodically shooting potato plant images through high-definition cameras installed above and on the side of the potato cultivation area; the environmental monitoring data is obtained by real-time monitoring the temperature parameters, light intensity parameters and carbon dioxide parameters and other related environmental parameters around the potato plant through temperature sensors, light intensity sensors and carbon dioxide sensors installed on the potato plant, so as to further understand the growth environment of the potato; the nutrient solution monitoring data is obtained by monitoring the content of each component in the nutrient solution through a plurality of high-precision sensors configured in the circulating pipeline, wherein the ion selective electrochemical sensor is used to accurately measure the ion concentrations of main nutrient elements such as nitrogen (N), phosphorus (P), potassium (K), calcium (Ca), magnesium (Mg) and sulfur (S), the biological sensor is used to detect the ion concentrations of trace elements such as iron (Fe), manganese (Mn), zinc (Zn) and copper (Cu), the pH sensor is used to monitor the pH of the nutrient solution, and the conductivity sensor is used to detect the total salt concentration of the nutrient solution to reflect the concentration level of the nutrient solution. The detection data of the above sensors are comprehensively analyzed to obtain the nutrient solution monitoring data. The technical solution provided in the present application accurately obtains multiple aspects of potato cultivation data through multiple ways, considers more comprehensive factors affecting the growth of potatoes, and improves the accuracy of the regulation scheme obtained based on the potato cultivation data.

[0027] In an embodiment of the present application, the scheme control module 02 includes a nutrient solution prediction unit and a scheme control unit. The nutrient solution prediction unit is used to analyze the growth monitoring data according to the CNN model, determine the potato growth stage, and output the growth stage one-hot encoding vector. In a specific embodiment, the collected growth monitoring data is first pre-processed. The growth monitoring data is image data. The image size is uniformly adjusted. The adjusted image size is 224 in length, 224 in width, and 3 in number of color channels. The image data is normalized so that the pixel value is scaled to the range of 0-1, and finally the processed growth monitoring data is obtained; furthermore, the processed growth monitoring data is input into the CNN model. The processed growth monitoring data is scanned by the underlying convolution kernels of different sizes in the CNN model. Optionally, the underlying convolution kernel includes 3 convolution layers, and the convolution kernel sizes are 3×3, 5×5, and 7×7, respectively. A sliding scan is performed on the processed growth monitoring data to identify the potato growth characteristic data. The potato growth characteristic data includes characteristic data such as the number of leaves, leaf color data, plant height data, and tuber morphology data of the plant. Secondly, the potato growth characteristic data is downsampled through the pooling layer (step size 2×2) to remove duplicate feature information and retain the core growth characteristic data. Furthermore, the core growth characteristic data are integrated through multi-scale fusion technology in the fully connected layer, and the core growth characteristic data are converted into the probability distribution of each growth stage through the normalized exponential function. The growth stages include germination stage, seedling stage, tuber formation stage, tuber swelling stage, and maturity stage. Finally, the growth stage with the largest probability value is used as the predicted current growth stage, and the probability distribution of the current growth stage is converted into a one-hot encoding vector of the growth stage.

[0028] Furthermore, the nutrient solution prediction unit is also used to combine environmental monitoring data, nutrient solution monitoring data and growth stage unique hot encoding vectors to obtain nutrient solution demand data according to the GRU model. In a specific embodiment, first, each growth stage is encoded into a five-dimensional growth stage vector, the environmental monitoring data is encoded into a four-dimensional environmental data vector, and the nutrient solution monitoring data is encoded into a twelve-dimensional nutrient solution data vector. Optionally, the five-dimensional growth stage vector includes a germination vector, a seedling stage vector, a tuber formation stage vector, a tuber swelling stage vector and a maturity stage vector. The four-dimensional environmental data vector includes a temperature and humidity vector, a carbon dioxide concentration vector and a light intensity vector. The twelve-dimensional nutrient solution data vector includes an N ion concentration vector, a P ion concentration vector, a K ion concentration vector, a Ca ion concentration vector, a Mg ion concentration vector, a S ion concentration vector, a Fe ion concentration vector, Mn ion concentration vector, Zn ion concentration vector, Cu ion concentration vector, pH vector and conductivity vector; concatenate the five-dimensional growth stage vector, the four-dimensional environmental data vector and the twelve-dimensional nutrient solution data vector to obtain a twenty-one-dimensional feature vector; construct the twenty-one-dimensional feature vector into a time series input sequence according to the daily time (24 hours) step, the time series input sequence includes the number of samples, time step and feature dimension, the time step is 24, and the feature dimension is 21; input the time series input sequence into the GRU model, and for each time step, use the logic function in the update gate of the GRU model to calculate the linear combination value of the current input sequence and the historical hidden state sequence , dynamically retain historical valid information, that is, dynamically determine how much valid information in the historical hidden state is retained. One thing that needs to be explained is that the update gate will retain as much historical valid information as possible to maintain the growth trend. For example, a large amount of potassium is required during the tuber enlargement period. The update gate will give a higher weight to the high potassium formula stage to maintain the nutrient supply trend; in the reset gate of the GRU model, the weight of historical information to be ignored in the historical valid information is screened and calculated, and the response abnormal data is calculated to control the degree of fusion of the current input sequence and the historical state sequence. For example, when the ion concentration is abnormal, when the conductivity value is detected to be abnormally high (ie, salt accumulation), the reset gate weakens the historical state. The memory of the solution formula forces the model to focus on the current ion balance requirements, combines the historical information filtered by the reset gate with the current input sequence, and generates a candidate hidden state sequence through activation of the hyperbolic tangent activation function; in the update gate, the candidate hidden state sequence is weightedly fused with the historical hidden state sequence to obtain the current hidden state sequence and pass it moment by moment. This mechanism enables the GRU model to dynamically adjust the degree of information retention and update at each time step, thereby better capturing the long-term dependencies and dynamic change characteristics in the time series; finally, the hidden state sequence of the final time step is mapped through the fully connected layer to obtain the nutrient solution demand data for the current growth stage.

[0029] In an embodiment of the present application, the scheme control unit is used to fuse the nutrient solution demand data and the nutrient solution monitoring data according to the DRL model to generate a dynamic nutrient solution control strategy. In a specific embodiment, in the DRL model, a state space, an action space and a reward function are constructed based on the nutrient solution demand data, the nutrient solution monitoring data and the growth stage unique hot encoding vector. The state space includes the actual nutrient solution monitoring data and the nutrient solution demand data predicted by the GRU model. For example, the current nutrient solution monitoring data is a twelve-dimensional nutrient solution data vector (N, P, K, Ca, Mg, S, Fe, Mn, Zn, Cu, pH and conductivity), the predicted nutrient solution demand data is the same twelve-dimensional nutrient solution data vector as the current nutrient solution monitoring data, the growth stage is a five-dimensional growth stage vector (germination period, seedling period, tuber formation period, tuber swelling period and maturity period), the action space is the parameter adjustment amount of the elements in the nutrient solution, for example, the N adjustment amount range is ±20% of the current value, and the calculation formula of the reward function is expressed by the following formula (1): Formula (1); Where, represents the reward function; Indicates the difference between the actual adjustment amount and the target adjustment amount; Indicates the nutrient solution requirement data, used to ensure that core elements (N / P / K, etc.) approach the target first; Indicates the nutrient concentration matching reward; Indicates the difference in pH before and after regulation, used to suppress drastic pH fluctuations; Indicates the difference between actual and target conductivity; Indicates the target conductivity during the growth stage (e.g. the target conductivity during tuber bulking is 2.5 mS / cm); Indicates conductivity concentration stability bonus; Indicates the pump running time, with linear deduction based on the total running time (seconds) of the mother liquid pump to encourage efficient and short-time operation; 、 、 、 They represent the weight coefficients of nutrient concentration matching reward, pH stability reward, conductivity concentration stability reward and pump running time respectively.

[0030] Furthermore, the DRL model obtains a dynamic nutrient solution control strategy based on the Actor-Critic framework. Specifically, in the Actor network, it receives the state space and outputs the action space, that is, outputs a specific nutrient solution control action, such as adding 30 mg / L of potassium, executes the action space and obtains the immediate nutrient solution parameters, and calculates the reward value in combination with the reward function; in the Critic network, it calculates the evaluation value between the state space and the action space, calculates the value error based on the reward function and the temporal difference algorithm, and updates the Critic network parameters based on the value error; calculates the gradient direction of the evaluation value, inputs the gradient direction of the evaluation value to the Actor network through the Critic network, updates the Actor network parameters, and obtains a dynamic nutrient solution control strategy. The technical solution provided by the embodiments of the present application, during the DRL model training process, stores historical state-action-reward samples through a historical database, combines the target network soft update to stabilize the learning process, uses reward feedback (such as the reward function of the nutrient solution concentration approaching the predicted value and the stability reward) to drive the dynamic adjustment of the strategy, and continuously repeats the "execute action-obtain feedback-update network" cycle, so that the model learns the optimal nutrient solution control strategy for different growth stages through continuous trial and error, and ultimately achieves accurate and dynamic nutrient solution supply control.

[0031] The technical solution provided in this application combines potato cultivation data including growth monitoring data, environmental monitoring data and nutrient solution monitoring data, and processes the potato cultivation data accordingly through the CNN-GRU model to obtain nutrient solution demand data, thereby accurately predicting the nutrient solution demand of potatoes at different growth stages. The dynamic nutrient solution regulation strategy is obtained through the DRL model, so that the nutrient solution supply is more in line with the actual growth needs, avoiding excess or deficiency of nutrition, thereby promoting the healthy growth of potatoes and improving potato yield and quality.

[0032] In the embodiment of the present application, the solution execution optimization module 03 includes a multi-liquid storage tank collaborative control unit, an environmental auxiliary control unit and a feedback correction unit. The multi-liquid storage tank collaborative control unit is used to control the nutrient solution formula through a pulse width modulation metering pump according to the dynamic nutrient solution control strategy. Specifically, 8 independent mother liquid storage tanks with a capacity of 50-200L are configured. The mother liquid storage tanks are built with different basic nutrients (such as potassium nitrate and diammonium phosphate). The mother liquid is mixed in proportion through a pulse width modulation metering pump (accuracy ±0.5%) to control the nutrient solution formula and achieve 0.1mL level precise ratio. For example, when the intelligent decision-making layer determines that potassium ions and calcium ions need to be supplemented at the same time, the system gives priority to potassium nitrate and calcium nitrate mother liquids to avoid the use of calcium salts containing sulfate to cause precipitation, thereby ensuring the stability of the nutrient solution.

[0033] Furthermore, the environmental auxiliary control unit is used to control the environmental monitoring data through the supplemental lighting equipment, ventilation equipment, and carbon dioxide control equipment according to the dynamic nutrient solution control strategy. Specifically, the multiple devices are linked and work together. The supplemental lighting system adjusts the light intensity based on the light sensor data and the potato growth stage. For example, during the tuber formation period, the proportion of red light in the LED spectrum is automatically increased to 60%, and the lighting duration is extended to 16 hours / day to promote starch synthesis. The ventilation equipment monitors the wind speed in the environment in real time. If the wind speed exceeds 3m / s, the top vents are automatically closed to prevent the potato plants from falling due to excessive wind. The carbon dioxide control device monitors the carbon dioxide concentration in the environment in real time. When the carbon dioxide concentration falls below 800ppm, the carbon dioxide generator is immediately activated to replenish the gas.

[0034] Furthermore, the feedback correction unit is configured to calculate the potato growth rate based on the regulated potato growth parameters, calculate a nutrient solution parameter correction based on the regulated nutrient solution parameters and a proportional-integral-differential control algorithm, and optimize the nutrient solution formula based on the potato growth rate and the nutrient solution parameter corrections. In a specific embodiment, feedback correction is accomplished via a rapid regulation loop and a long-term optimization loop. The rapid regulation loop is centered around a pH / conductivity sensor. After each nutrient solution regulation, the sensor collects the regulated nutrient solution parameters once per second and transmits them to the feedback correction unit. The feedback correction unit then calculates the nutrient solution parameter corrections in real time based on the proportional-integral-differential control algorithm (with a proportional coefficient of 0.5-2.0 and an integration time of 10-30 minutes) and the regulated nutrient solution parameters, and optimizes the nutrient solution formula based on the nutrient solution parameter corrections. For example, if the conductivity value exceeds the target by 5% after regulation, the system immediately fine-tunes the mother liquor mixing ratio to ensure that conductivity fluctuations are controlled within ±5% and pH fluctuations are within ±0.2, achieving rapid and accurate correction. The long-term optimization loop obtains regulated potato growth parameters weekly, including growth parameters such as leaf area and tuber volume. Based on these regulated potato growth parameters, it calculates the potato growth rate, including leaf area growth rate and tuber volume increment, and optimizes the nutrient solution formula based on the potato growth rate. In other words, if the actual potato growth rate deviates from the expected potato growth rate, the nutrient solution formula problem is automatically traced, triggering a GRU model parameter update and the DRL model to re-optimize the nutrient solution regulation strategy. For example, if the tuber volume increment is less than 10% of the model's predicted increment, the formula optimization is triggered, and the DRL model re-outputs the optimized nutrient solution regulation strategy. The technical solution provided in this application uses a rapid adjustment loop to complete corrections within one hour after nutrient solution control, and a long-term optimization loop to complete growth rate assessment and feedback optimization every Sunday morning. Through this full-cycle closed-loop iteration of "data collection - solution execution - effect feedback - system optimization," the solution execution optimization module not only accurately implements the dynamic nutrient solution control strategy output by the solution control module, but also continuously iterates to optimize the control strategy, making it adaptable to the different growth stages and growth environments of potatoes. This closed-loop feedback correction mechanism can improve the accuracy and sensitivity of nutrient solution parameter control. Furthermore, potato cultivation data detection and nutrient solution control are all automatically completed by the system, eliminating the need for frequent manual operation and monitoring, greatly reducing labor costs, and thus achieving precision, intelligence, and efficiency in nutrient solution-controlled potato cultivation.

[0035] For a specific embodiment, please refer to Figure 2 , which shows a detailed schematic diagram of a potato soilless cultivation system based on multi-component nutrient solution control provided in an embodiment of the present application, such as Figure 2As shown, the system includes a data monitoring module, a scheme control module and a scheme execution optimization module. The data monitoring module includes the collected potato cultivation data including growth monitoring data, environmental monitoring data and nutrient solution monitoring data. The scheme control module includes a nutrient solution prediction unit and a scheme control unit. In the nutrient solution prediction unit, the growth monitoring data in the data monitoring module is analyzed according to the CNN model to determine the potato growth stage and output a one-hot encoding vector of the growth stage. The environmental monitoring data, nutrient solution monitoring data and the one-hot encoding vector of the growth stage in the data monitoring module are combined to obtain the nutrient solution demand data according to the gated cyclic unit model. In the scheme control unit, the nutrient solution demand data obtained by the nutrient solution prediction unit and the nutrient solution monitoring data in the data monitoring module are integrated according to the DRL model to generate a dynamic nutrient solution control strategy. The scheme execution optimization module includes a multi-liquid storage tank collaborative control unit, an environmental auxiliary control unit and a feedback correction unit. In the multi-liquid storage tank collaborative control unit, the dynamic nutrient solution control strategy obtained by the scheme control unit is used to control the nutrient solution formula in the data monitoring module through the pulse width modulation metering pump. In the environmental auxiliary control unit, the dynamic nutrient solution control strategy obtained by the scheme control unit is used to control the environmental monitoring data in the data monitoring module through the supplementary lighting equipment, ventilation equipment and carbon dioxide control equipment. In the feedback correction unit, the potato growth rate is calculated according to the regulated potato growth parameters. The nutrient solution parameter correction amount is calculated according to the regulated nutrient solution parameters and the proportional-integral-differential control algorithm. The nutrient solution control strategy obtained by the scheme control module is optimized according to the potato growth rate and the nutrient solution parameter correction amount, thereby optimizing the nutrient solution formula.

[0036] In summary, the embodiment of the present application provides a potato soilless cultivation system based on multi-component nutrient solution control, which monitors and collects potato cultivation data in real time through a data monitoring module. The potato cultivation data includes growth monitoring data, environmental monitoring data and nutrient solution monitoring data. Various aspects of potato cultivation data are accurately obtained through multiple methods, and the factors affecting potato growth are considered more comprehensively, thereby improving the accuracy of the subsequent control scheme obtained based on the potato cultivation data; through the scheme control module, according to the CNN-GRU model, the nutrient solution demand data is obtained in combination with the potato cultivation data, so as to achieve accurate prediction of the nutrient solution demand of potatoes at different growth stages; according to the DRL model, the nutrient solution demand data and the nutrient solution monitoring data are combined to generate a dynamic nutrient solution control strategy, which is convenient for flexible adjustment according to the potato cultivation conditions; through the scheme execution optimization The module executes dynamic nutrient solution control strategy, regulates environmental monitoring data and nutrient solution formula, so that the nutrient solution supply and environmental conditions are more in line with the actual growth needs of potatoes, avoiding problems such as overnutrition, undernutrition and unsuitable environment, thereby promoting the healthy growth of potatoes and improving potato yield and quality. In addition, the nutrient solution formula is optimized according to the nutrient solution parameters and potato growth parameters. Through this full-cycle closed-loop iteration of "data collection-execution plan-effect feedback-system optimization", the control strategy is accurately implemented, and it is continuously iterated to optimize the control strategy so that the control strategy can adapt to the different growth stages and growth environment of potatoes. Through this closed-loop feedback correction mechanism, the accuracy and sensitivity of nutrient solution parameter control can be improved, and labor costs are greatly reduced, thereby realizing the precision, intelligence and efficiency of potato cultivation controlled by nutrient solution.

[0037] The above is an introduction to the system embodiment of the present application. Based on the aforementioned embodiment, the method embodiment of the present application is introduced below.

[0038] Please refer to Figure 3 , which shows a flow chart of a potato soilless cultivation method based on multi-component nutrient solution control provided by an embodiment of the present application, which is applied to Figure 1 The potato soilless cultivation system based on multi-component nutrient solution control is shown. For details not disclosed in the method embodiment, please refer to the system embodiment. The system includes a data monitoring module, a program control module and a program execution optimization module. The program execution optimization module is connected to the monitoring module and the program control module respectively, and the data monitoring module, the program control module and the program execution optimization module are connected in sequence. Figure 3 As shown, the startup method includes the following steps S310 to S330.

[0039] Step S310 , real-time monitoring and collection of potato cultivation data, wherein the potato cultivation data includes growth monitoring data, environment monitoring data, and nutrient solution monitoring data.

[0040] In an embodiment of the present application, the growth monitoring data is obtained by regular photography by cameras placed above and on the sides of the potato cultivation area; the environmental monitoring data is obtained by real-time monitoring by temperature sensors, light intensity sensors and carbon dioxide sensors installed on the potato plants; the nutrient solution monitoring data is obtained by monitoring by high-precision sensors in the circulation pipeline, and the high-precision sensors include ion-selective electrobiosensors, pH sensors and conductivity sensors.

[0041] Step S320: According to the convolutional neural network-gated recurrent unit model, combined with the potato cultivation data, nutrient solution demand data is obtained, and according to the deep reinforcement learning model, combined with the nutrient solution demand data and the nutrient solution monitoring data, a dynamic nutrient solution control strategy is generated.

[0042] In an embodiment of the present application, the growth monitoring data is analyzed according to the CNN model to determine the potato growth stage, and a one-hot encoding vector of the growth stage is output. The environmental monitoring data, the nutrient solution monitoring data, and the one-hot encoding vector of the growth stage are combined to obtain the nutrient solution demand data according to the gated recurrent unit model; the nutrient solution demand data and the nutrient solution monitoring data are fused according to the DRL model to generate the dynamic nutrient solution control strategy; In an embodiment of the present application, the growth monitoring data is preprocessed and normalized to obtain processed growth monitoring data; the processed growth monitoring data is input into the CNN model, and the processed growth monitoring data is scanned by the bottom convolution kernel to identify potato growth characteristic data; the potato growth characteristic data is downsampled by the pooling layer to remove repeated feature information and retain the core growth characteristic data; the core growth characteristic data is converted into the probability distribution of each growth stage through multi-scale fusion technology and normalized exponential function, and each growth stage includes germination stage, seedling stage, tuber formation stage, tuber swelling stage and maturity stage; the growth stage with the largest probability value is taken as the current growth stage, and the probability distribution of the current growth stage is converted into the one-hot encoding vector of the growth stage.

[0043] In an embodiment of the present application, each growth stage, the environmental monitoring data and the nutrient solution monitoring data are respectively encoded into a five-dimensional growth stage vector, a four-dimensional environmental data vector and a twelve-dimensional nutrient solution data vector, and the five-dimensional growth stage vector, the four-dimensional environmental data vector and the twelve-dimensional nutrient solution data vector are spliced ​​to obtain a twenty-one-dimensional feature vector; the twenty-one-dimensional feature vector is constructed as a time series input sequence according to the daily time step, and the time series input sequence includes the number of samples, the time step and the feature dimension, the time step is 24, and the feature dimension is 21; the time series input sequence is input into the gated recurrent unit model, and in the update gate, the Sigmoid function is used to calculate the linear combination value of the current input sequence and the historical hidden state sequence, and the historical valid information is dynamically retained; in the reset gate, the historical information weights and response abnormal data to be ignored in the historical valid information are screened and calculated, and the historical information is combined with the current input sequence, and the historical information is combined through HYPERLINK "https: / / baike.baidu.com / item / %E5%8F%8C%E6%9B%B2%E6%AD%A3%E5%88%87%E5%87%BD%E6%95%B0 / 15469414" \t "_blank"The hyperbolic tangent function (tanh) is activated to generate a candidate hidden state sequence; in the update gate, the candidate hidden state sequence is weightedly fused with the historical hidden state sequence to obtain the current hidden state sequence; the hidden state sequence of the final time step is mapped through the fully connected layer to obtain the nutrient solution demand data for the current growth stage.

[0044] In an embodiment of the present application, in the DRL model, a state space, an action space, and a reward function are constructed according to the nutrient solution demand data, the nutrient solution monitoring data, and the growth stage one-hot encoding vector. The calculation formula of the reward function is expressed by the following formula: ; Where, represents the reward function; Indicates the difference between the actual adjustment amount and the target adjustment amount; Indicates the nutrient solution demand data; Indicates the nutrient concentration matching reward; Indicates the difference in pH before and after regulation; Indicates the difference between actual and target conductivity; represents the target conductivity during the growth phase; Indicates conductivity concentration stability bonus; Indicates the pump running time; 、 、 、 They respectively represent the weight coefficients of the nutrient concentration matching reward, the pH stability reward, the conductivity concentration stability reward and the pump running time; in the Actor network, the state space is received and the action space is output, the action space is executed and the real-time nutrient solution parameters are obtained, and the reward value is calculated in combination with the reward function; in the Critic network, the evaluation value between the state space and the action space is calculated, the value error is calculated according to the reward function and the temporal difference algorithm, and the Critic network parameters are updated according to the value error; the gradient direction of the evaluation value is input to the Actor network through the Critic network, the Actor network parameters are updated, and the dynamic nutrient solution control strategy is obtained.

[0045] Step S330 , executing the dynamic nutrient solution control strategy, controlling the environmental monitoring data and the nutrient solution formula, and optimizing the nutrient solution formula according to the nutrient solution parameters and potato growth parameters.

[0046] In an embodiment of the present application, according to the dynamic nutrient solution control strategy, the nutrient solution formula is controlled by a pulse width modulation metering pump; according to the dynamic nutrient solution control strategy, the environmental monitoring data is controlled by a lighting device, a ventilation device, and a carbon dioxide control device; the potato growth rate is calculated based on the regulated potato growth parameters, the nutrient solution parameter correction amount is calculated based on the regulated nutrient solution parameters and a proportional-integral-differential control algorithm, and the nutrient solution formula is optimized based on the potato growth rate and the nutrient solution parameter correction amount.

[0047] In summary, the embodiment of the present application provides a soilless potato cultivation method based on multi-component nutrient solution control, which monitors and collects potato cultivation data in real time. The potato cultivation data includes growth monitoring data, environmental monitoring data and nutrient solution monitoring data. The potato cultivation data of various aspects is accurately obtained through multiple methods, and the factors affecting potato growth are considered more comprehensively, thereby improving the accuracy of the subsequent control scheme obtained based on the potato cultivation data; according to the CNN-GRU model, the nutrient solution demand data is obtained in combination with the potato cultivation data, so as to realize the accurate prediction of the nutrient solution demand of potatoes at different growth stages; according to the DRL model, the nutrient solution demand data and the nutrient solution monitoring data are combined to generate a dynamic nutrient solution control strategy, which is convenient for flexible adjustment according to the potato cultivation conditions; the dynamic nutrient solution control strategy is executed The system uses a closed-loop control strategy to adjust environmental monitoring data and nutrient solution formula to make the nutrient solution supply and environmental conditions more in line with the actual growth needs of potatoes, avoid problems such as overnutrition, undernutrition and unsuitable environment, thereby promoting the healthy growth of potatoes and improving potato yield and quality. In addition, the nutrient solution formula is optimized according to the nutrient solution parameters and potato growth parameters. Through this full-cycle closed-loop iteration of "data collection-implementation plan-effect feedback-system optimization", the control strategy is accurately implemented, and it is continuously iterated to optimize the control strategy so that the control strategy can adapt to the different growth stages and growth environment of potatoes. Through this closed-loop feedback correction mechanism, the accuracy and sensitivity of nutrient solution parameter control can be improved, and labor costs can be greatly reduced, thereby realizing the precision, intelligence and efficiency of potato cultivation controlled by nutrient solution.

[0048] It should be noted that in the embodiments of the present application, if the aforementioned soilless potato cultivation method based on multi-component nutrient solution control is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiments of the present application, or the portion that contributes to the relevant technology, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing an electronic device to execute all or part of the methods described in each embodiment of the present application. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, a mobile hard drive, a read-only memory (ROM), a magnetic disk, or an optical disk. Thus, the embodiments of the present application are not limited to any specific combination of hardware and software.

[0049] Correspondingly, the embodiment of the present application provides a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to implement the steps of any one of the soilless cultivation methods of potato based on multi-component nutrient solution control. Correspondingly, the embodiment of the present application also provides a computer program product, which is executed by a processor of an electronic device to implement the steps of any one of the soilless cultivation methods of potato based on multi-component nutrient solution control.

[0050] Based on the same technical concept, the embodiment of the present application provides an electronic device for implementing the soilless cultivation method of potato based on multi-component nutrient solution control. Figure 4 The hardware entity diagram of the electronic device provided by the embodiment of the present application is shown in FIG. 4. Figure 4 As shown in FIG. 4, the electronic device 400 includes a memory 410 and a processor 420. The memory 410 stores a computer program executable by the processor 420. The processor 420 implements the steps of any one of the soilless cultivation methods of potato based on multi-component nutrient solution control when executing the program.

[0051] The memory 410 is configured to store instructions and applications executable by the processor 420, and can also cache data to be processed by the processor 420 and each module in the electronic device (for example, image data, audio data, voice communication data and video communication data), which can be implemented by FLASH or RAM.

[0052] The processor 420 implements the steps of any one of the soilless cultivation methods of potato based on multi-component nutrient solution control when executing the program. The processor 420 generally controls the overall operation of the electronic device 400.

[0053] The processor may be at least one of an application-specific integrated circuit (ASIC), a digital signal processor (DSP), a digital signal processing device (DSPD), a programmable logic device (PLD), a field programmable gate array (FPGA), a central processing unit (CPU), a controller, a microcontroller, and a microprocessor. It is understood that the electronic device that implements the functions of the processor may also be other electronic devices, and the embodiments of the present application are not specifically limited thereto.

[0054] The above-mentioned computer storage medium / memory can be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), a magnetic random access memory (FRAM), a flash memory (Flash Memory), a magnetic surface storage device, an optical disc, or a compact disc read-only memory (CD-ROM); it can also be various electronic devices that include one or any combination of the above-mentioned memories, such as mobile phones, computers, tablet devices, personal digital assistants, etc.

[0055] It should be noted that the description of the above storage medium and device embodiments is similar to the description of the above method embodiments and has similar beneficial effects as the method embodiments. For technical details not disclosed in the storage medium and device embodiments of this application, please refer to the description of the method embodiments of this application for understanding.

[0056] It should be understood that "one embodiment" or "an embodiment" mentioned throughout the specification means that the specific features, structures or characteristics related to the embodiment are included in at least one embodiment of the present application. Therefore, "in one embodiment" or "in an embodiment" appearing throughout the specification does not necessarily refer to the same embodiment. In addition, these specific features, structures or characteristics can be combined in one or more embodiments in any suitable manner. It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application. The above-mentioned serial numbers of the embodiments of the present application are for description only and do not represent the advantages and disadvantages of the embodiments.

[0057] It should be noted that, in this document, the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, article, or apparatus comprising a series of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, method, article, or apparatus comprising the element.

[0058] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely schematic. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as: multiple units or components can be combined, or can be integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the components shown or discussed can be through some interfaces, and the indirect coupling or communication connection of the devices or units can be electrical, mechanical or other forms.

[0059] The units described above as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units; they may be located in one place or distributed across multiple network units; some or all of the units may be selected according to actual needs to achieve the purpose of the embodiment of the present application.

[0060] In addition, all functional units in the embodiments of the present application can be integrated into one processing unit, or each unit can be a separate unit, or two or more units can be integrated into one unit; the above-mentioned integrated units can be implemented in the form of hardware or in the form of hardware plus software functional units.

[0061] Alternatively, if the above-mentioned integrated unit of the present application is implemented in the form of a software function module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiment of the present application, or the part that contributes to the relevant technology, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling the automatic test line of the device to execute all or part of the methods described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as mobile storage devices, ROMs, magnetic disks or optical disks.

[0062] The methods disclosed in the several method embodiments provided in this application can be arbitrarily combined without conflict to obtain new method embodiments.

[0063] The features disclosed in the several method or device embodiments provided in this application can be arbitrarily combined without conflict to obtain new method embodiments or device embodiments.

[0064] The above is merely an embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

Claims

1. A potato soilless cultivation system based on multi-component nutrient solution control, characterized in that: The system comprises: A data monitoring module is used to monitor and collect potato cultivation data in real time, wherein the potato cultivation data includes growth monitoring data, environmental monitoring data and nutrient solution monitoring data; a scheme control module for obtaining nutrient solution demand data based on a convolutional neural network-gated recurrent unit model in combination with the potato cultivation data, and generating a dynamic nutrient solution control strategy based on a deep reinforcement learning model in combination with the nutrient solution demand data and the nutrient solution monitoring data; The program execution optimization module is used to execute the dynamic nutrient solution control strategy, control the environmental monitoring data and the nutrient solution formula, and optimize the nutrient solution formula according to the nutrient solution parameters and potato growth parameters.

2. The system according to claim 1, wherein: The scheme control module includes a nutrient solution prediction unit and a scheme control unit, wherein: The nutrient solution prediction unit is used to analyze the growth monitoring data according to a convolutional neural network model, determine the potato growth stage, output a one-hot encoding vector of the growth stage, and combine the environmental monitoring data, the nutrient solution monitoring data, and the one-hot encoding vector of the growth stage to obtain the nutrient solution demand data according to a gated recurrent unit model; The scheme control unit is used to fuse the nutrient solution demand data and the nutrient solution monitoring data according to a deep reinforcement learning model to generate the dynamic nutrient solution control strategy.

3. The system according to claim 2, characterized in that The step of analyzing the growth monitoring data according to the convolutional neural network model, determining the potato growth stage, and outputting a one-hot encoding vector of the growth stage includes: performing preprocessing and normalization processing on the growth monitoring data to obtain processed growth monitoring data; Inputting the processed growth monitoring data into the convolutional neural network model, scanning the processed growth monitoring data through the underlying convolution kernel to identify potato growth characteristic data; Downsampling the potato growth characteristic data through a pooling layer to remove duplicate feature information and retain core growth characteristic data; The core growth characteristic data are converted into probability distributions of each growth stage through multi-scale fusion technology and normalized exponential function, and each growth stage includes germination stage, seedling stage, tuber formation stage, tuber expansion stage and maturity stage; The growth stage with the largest probability value is taken as the current growth stage, and the probability distribution of the current growth stage is converted into the one-hot encoding vector of the growth stage.

4. The system according to claim 2, wherein: The combining of the environmental monitoring data, the nutrient solution monitoring data, and the growth stage one-hot encoding vector to obtain the nutrient solution demand data according to a gated recurrent unit model includes: Encoding each growth stage, the environmental monitoring data, and the nutrient solution monitoring data into a five-dimensional growth stage vector, a four-dimensional environmental data vector, and a twelve-dimensional nutrient solution data vector, respectively, and concatenating the five-dimensional growth stage vector, the four-dimensional environmental data vector, and the twelve-dimensional nutrient solution data vector to obtain a twenty-one-dimensional feature vector; Constructing the twenty-one-dimensional feature vector into a time series input sequence according to the daily time step, wherein the time series input sequence includes the number of samples, the time step and the feature dimension, wherein the time step is 24 and the feature dimension is 21; Input the time series input sequence into the gated recurrent unit model, and in the update gate, use the logic function to calculate the linear combination value of the current input sequence and the historical hidden state sequence to dynamically retain historical valid information; In the reset gate, the weight of historical information to be ignored in the historical valid information and the response abnormal data are screened and calculated, the historical information is combined with the current input sequence, and the candidate hidden state sequence is generated through the hyperbolic tangent activation function. In the update gate, the candidate hidden state sequence is weightedly fused with the historical hidden state sequence to obtain the current hidden state sequence; The hidden state sequence of the final time step is mapped through the fully connected layer to obtain the nutrient solution requirement data of the current growth stage.

5. The system according to claim 2, wherein: The step of fusing the nutrient solution demand data and the nutrient solution monitoring data according to the deep reinforcement learning model to generate the dynamic nutrient solution regulation strategy includes: In the deep reinforcement learning model, the state space, action space and reward function are constructed according to the nutrient solution demand data, the nutrient solution monitoring data and the growth stage one-hot encoding vector. The calculation formula of the reward function is expressed by the following formula: ； Where, represents the reward function; Indicates the difference between the actual adjustment amount and the target adjustment amount; Indicates the nutrient solution demand data; Indicates the nutrient concentration matching reward; Indicates the difference in pH before and after regulation; Indicates the difference between actual and target conductivity; represents the target conductivity during the growth phase; Indicates conductivity concentration stability bonus; Indicates the pump running time; 、 、 、 They represent the weight coefficients of nutrient concentration matching bonus, pH stability bonus, conductivity concentration stability bonus and pump running time respectively; In the executor network, the state space is received and the action space is output, the action space is executed and the instant nutrient solution parameters are obtained, and the reward value is calculated in combination with the reward function; In the evaluator network, the evaluation value between the state space and the action space is calculated, the value error is calculated according to the reward function and the temporal difference algorithm, and the critic network parameters are updated according to the value error; The gradient direction of the evaluation value is input into the executor network through the evaluator network, the executor network parameters are updated, and the dynamic nutrient solution control strategy is obtained.

6. The system according to claim 1, wherein: The solution execution optimization module includes a multi-tank collaborative control unit, an environmental auxiliary control unit, and a feedback correction unit, wherein: The multi-liquid storage tank coordinated control unit is used to control the nutrient solution formula through a pulse width modulation metering pump according to the dynamic nutrient solution control strategy; The auxiliary environmental control unit is used to control the environmental monitoring data through the lighting equipment, ventilation equipment and carbon dioxide control equipment according to the dynamic nutrient solution control strategy; The feedback correction unit is used to calculate the potato growth rate based on the regulated potato growth parameters, calculate the nutrient solution parameter correction amount based on the regulated nutrient solution parameters and the proportional-integral-differential control algorithm, and optimize the nutrient solution formula based on the potato growth rate and the nutrient solution parameter correction amount.

7. The system according to claim 1, wherein: The growth monitoring data is obtained by regularly photographing the potatoes using cameras placed above and to the sides of the potato cultivation area; The environmental monitoring data is obtained by real-time monitoring through temperature sensors, light intensity sensors and carbon dioxide sensors installed on potato plants; The nutrient solution monitoring data is obtained by monitoring high-precision sensors in the circulation pipeline, and the high-precision sensors include ion-selective electrical biosensors, pH sensors, and conductivity sensors.

8. A potato soilless cultivation method based on multi-component nutrient solution control, characterized in that: The invention is applied to a potato soilless cultivation system based on multi-component nutrient solution control, wherein the system includes a data monitoring module, a program control module, and a program execution optimization module. The method includes: Real-time monitoring and collection of potato cultivation data, including growth monitoring data, environmental monitoring data, and nutrient solution monitoring data; According to the convolutional neural network-gated recurrent unit model, combined with the potato cultivation data, nutrient solution demand data is obtained, and according to the deep reinforcement learning model, combined with the nutrient solution demand data and the nutrient solution monitoring data, a dynamic nutrient solution control strategy is generated; The dynamic nutrient solution regulation strategy is executed to regulate the environmental monitoring data and the nutrient solution formula, and the nutrient solution formula is optimized according to the nutrient solution parameters and potato growth parameters.

9. An electronic device comprising a memory and a processor, wherein the memory stores a computer program that can be run on the processor, wherein: When the processor executes the program, the steps in the method of claim 8 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to claim 8 are implemented.

Images