Measurement method and system for nutrient solution components of hydroponic plant

Abstract

A measurement method and system for nutrient solution components of a hydroponic plant, which method and system relate to the technical field of plant nutrition management and hydroponic cultivation. The measurement method comprises: setting growth node time periods for a plant, performing, multiple times at set time points, nutrient component measurement on a nutrient solution where the plant is located, and recording water consumption amounts; using a neural network to establish a detection model by means of the acquired water consumption amounts and data of a nutrient component growth curve, training the detection model, and storing the optimal model; and measuring nutrient components at different time points in real time, and when the difference between an actual measured value and a predicted value of the optimal model is greater than a set error threshold value, triggering a replenishment and adjustment mechanism to ensure that components of the nutrient solution are always maintained within the optimal range. Thus, the growth speed and quality of plants are improved, the frequency and difficulty of manual intervention are reduced, measurement is accurately performed on the basis of a corresponding plant growth period, and a nutrient solution ratio is adjusted on the basis of a measurement result, so as to meet the growth requirements of the plants.

Description

Methods and systems for detecting components in hydroponic plant nutrient solutions Technical Field

[0001] This invention belongs to the field of plant nutrient management and hydroponics technology, and particularly relates to a method and system for detecting the components of nutrient solutions for hydroponic plants. Background Technology

[0002] With the development of modern agriculture, hydroponics has become widely used as an efficient and environmentally friendly method of plant cultivation. However, the management of nutrient solution components is crucial in hydroponics. Traditional methods for detecting nutrient solution components usually require manual sampling and laboratory analysis, which is cumbersome and time-consuming, and cannot reflect the plant's growth needs and nutritional status in real time.

[0003] Different plants at different growth stages consume varying amounts of nutrient solution, and even the consumption of the same nutrients within the solution can differ. Different hydroponic methods have different advantages and disadvantages, but all methods require ensuring sufficient nutrients, pH levels, oxygen concentration, and other environmental conditions in the nutrient solution. Hydroponically grown plants require frequent nutrient solution changes. Even within the same variety, different plants exhibit varying nutrient absorption rates, resulting in different nutrient solution replacement cycles for each plant.

[0004] However, current tissue culture methods involve manually or mechanically replacing the nutrient solution at a fixed time. This results in some plants that consume more nutrients facing nutrient deficiency, while plants that consume nutrients more slowly are wasted. Therefore, how to achieve real-time and accurate detection of nutrient solution components has become an important issue in the development of hydroponic technology. Summary of the Invention

[0005] The main objective of this invention is to provide a method and system for detecting the components of hydroponic plant nutrient solutions, and to solve the technical problem of how to accurately detect the components according to the corresponding plant growth cycle and adjust the nutrient solution ratio according to the detection results to meet the plant's growth needs.

[0006] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is: a method for detecting the components of hydroponic plant nutrient solution, comprising the following steps:

[0007] S1: Obtain the dataset, set multiple growth node time periods for the plant, and within the set time interval of each growth node time period, detect the nutrient composition of the plant's nutrient solution and record the water consumption.

[0008] S2: The data obtained from the water consumption and nutrient composition are preprocessed and then divided into training and test sets according to a preset ratio.

[0009] S3: Use a neural network to build a detection model, use the training set to train the model, and save the optimal model;

[0010] S4: Real-time detection of nutrient components at different time points, using the optimal model for detection. When the difference between the concentration of nutrient components and the concentration predicted by the optimal model is greater than the set error threshold, nutrient solution is added, and the nutrient solution ratio is adjusted according to the concentration predicted by the optimal model.

[0011] In a preferred embodiment, S2 can also perform growth curve fitting, linearly fitting the obtained water consumption and nutrient composition to obtain the growth curves of the two with respect to time.

[0012] The S3 uses a neural network to build a detection model, uses a dataset of growth curves to train the model, and saves the optimal model corresponding to the growth curve.

[0013] In a preferred embodiment, S1 includes:

[0014] Plants are divided into multiple growth stages, including the budding stage, the growth stage, the flowering stage, and the fruiting stage.

[0015] Multiple preset time points are allowed. The nutrients in the nutrient solution are one or more of the following: nitrogen, phosphorus, potassium, calcium, iron, magnesium, sulfur, boron, zinc, copper, molybdenum and chlorine. The nutrients exist in the form of salts. The proportion of nutrients in each growth stage is determined based on the nutrient requirements of the plant at each growth stage.

[0016] Sensors are used to detect the ion concentration data of nutrients in the nutrient solution;

[0017] The daily water consumption of plants was recorded and correlated with the ion concentration data of nutrients.

[0018] In the preferred embodiment, S2 includes: linearly fitting the obtained water consumption and nutrient composition, using the least squares method for the fitting formula, wherein the nutrients are distributed proportionally, and some components are selected as characterization data, each group of nutrient composition characterization data is used as a feature value, and each feature value is linearly fitted with the corresponding water consumption; the characterization data are nitrogen, phosphorus, and potassium.

[0019] In the preferred embodiment, a detection model is established using a neural network to predict the relationship between the characterization data and the corresponding water consumption.

[0020] In the preferred embodiment, the model training utilizes iterative training to train the detection model, including:

[0021] Set the number of training iterations and input the training set into the detection model;

[0022] The detection model makes forward predictions;

[0023] The loss function is calculated based on the prediction results;

[0024] The obtained loss value is used for backpropagation to optimize the detection model;

[0025] If the current detection model is better than the previous detection model, save the optimal detection model.

[0026] In the preferred embodiment, adjusting the nutrient solution ratio according to the concentration predicted by the optimal model includes:

[0027] Based on the difference between the current nutrient concentration and the concentration predicted by the optimal model, the amount of nutrient solution and the amount of water output are controlled respectively. The total amount of nutrients required is calculated, and the nutrient solution is mixed with water to adjust the nutrient solution. The solution is replenished in multiple times and tested at fixed intervals. When the concentration predicted by the optimal model is reached, the adjustment is stopped and the water level in the plant planting pipeline is maintained.

[0028] In the preferred embodiment, the step of replenishing the nutrients multiple times and detecting them at fixed intervals specifically means: 4-6 times, with a fixed interval not less than the length of the plant planting pipeline divided by the current flow rate of the nutrient solution in the pot.

[0029] In the preferred embodiment, the power of the nutrient solution pump and the first water pump is adjusted, and after they are mixed in the mixing tank, the second water pump is started for irrigation.

[0030] In a preferred embodiment, the nutrient component detection in S1 also includes the detection of special elements, establishing a microcirculation branch based on the special elements required by the plant species, and detecting the concentration of the current special components.

[0031] A mobile detection system for components of hydroponic plant nutrient solution, comprising:

[0032] The acquisition module is used to set multiple growth node time periods for the plant. Within the set time interval of each growth node time period, the nutrient composition of the plant's nutrient solution is detected, and the water consumption is recorded at the same time.

[0033] The data preprocessing module is used to correlate the acquired water consumption and nutrient composition in the dataset, perform data preprocessing, and then divide the dataset into training and test sets according to a preset ratio.

[0034] The model building module is used to build a detection model using a neural network, train the model using a training set, and save the optimal model.

[0035] The prediction module is used to detect nutrient components at different time points in real time. It uses the optimal model for detection. When the difference between the concentration of the nutrient component and the concentration predicted by the optimal model is greater than the set error threshold, nutrient solution is added and the nutrient solution ratio is adjusted according to the concentration predicted by the optimal model.

[0036] This invention provides a method and system for detecting the components of nutrient solution for hydroponic plants. The method involves acquiring a dataset, mapping water consumption to nutrient components, preprocessing the data, and dividing it into a training set and a test set according to a preset ratio. A detection model is established using a neural network, and the optimal model is saved. Nutrient components are detected in real time at different time points using the optimal model. When the difference between the nutrient concentration and the concentration predicted by the optimal model exceeds a set error threshold, nutrient solution is replenished, and the nutrient solution ratio is adjusted according to the concentration predicted by the optimal model.

[0037] The beneficial effects of this invention are:

[0038] 1. By monitoring the composition of the nutrient solution in real time, we can understand the growth needs and nutritional status of plants in a timely manner, providing accurate data support for adjusting the nutrient solution;

[0039] 2. The use of neural networks to establish a detection model improves the accuracy and efficiency of detection, enables automatic adjustment of nutrient solution, and reduces the frequency and difficulty of manual intervention;

[0040] 3. By fitting growth curves, we can more accurately predict plant growth trends and nutrient requirements, providing a scientific basis for the long-term management of nutrient solutions. Attached Figure Description

[0041] The present invention will be further described below with reference to the accompanying drawings and embodiments:

[0042] Figure 1 is a flowchart of the detection method of the present invention;

[0043] Figure 2 is a schematic diagram of the detection system of the present invention. Detailed Implementation

[0044] Example 1

[0045] As shown in Figure 1, a method for detecting the components of a hydroponic plant nutrient solution includes the following steps:

[0046] S1: Obtain the dataset, set multiple growth node time periods for the plant, and within the set time interval of each growth node time period, detect the nutrient composition of the plant's nutrient solution and record the water consumption.

[0047] S2: The data obtained from the water consumption and nutrient composition are preprocessed and then divided into training and test sets according to a preset ratio.

[0048] S3: Use a neural network to build a detection model, use the training set to train the model, and save the optimal model.

[0049] S4: Real-time detection of nutrient components at different time points, using the optimal model for detection. When the difference between the concentration of nutrient components and the concentration predicted by the optimal model is greater than the set error threshold, nutrient solution is added, and the nutrient solution ratio is adjusted according to the concentration predicted by the optimal model.

[0050] In this embodiment, a dataset of relevant plant growth stages is acquired, and the acquired water consumption and nutrient composition are linearly fitted. A neural network is then used to build a detection model on the fitted curve, which is trained and the optimal model is saved. Finally, the nutrient composition at different time points is detected in real time, and the optimal model is used for detection. Based on the growth curve, this embodiment sets various time points for discontinuous continuous detection and builds a detection model. Finally, based on the growth curve model, the content of each component in the current plant nutrient solution can be obtained. Combined with the specific growth curve, the concentration of the nutrient solution and each component (such as nitrogen, phosphorus, and potassium) can be adjusted accordingly. This achieves intelligent and accurate detection of the nutrient composition of the nutrient solution, allowing for more precise replenishment and ratio, optimizing the composition of the hydroponic nutrient solution, improving plant growth, saving economic and labor costs, and reducing resource waste.

[0051] The data acquisition, modeling, and real-time monitoring and adjustment in this embodiment can help improve the growth efficiency and quality of hydroponic plants. It can accurately detect the growth cycle of the corresponding plants and accurately distribute the concentration according to the real-time concentration, thereby improving the planting efficiency of hydroponic plants and greatly enhancing economic benefits.

[0052] The nutrients in this embodiment include: 1. Nitrogen source: mainly urea and calcium nitrate, supplemented by potassium nitrate; 2. Phosphorus source: potassium dihydrogen phosphate and phosphoric acid are preferred; 3. Potassium source: mainly potassium sulfate, supplemented by potassium nitrate; 4. Calcium is provided by calcium nitrate; 5. Magnesium source: magnesium sulfate; Iron source: chelated iron; 6. Trace elements: copper, zinc, manganese, boron, molybdenum, and chloride have relatively stable chemical properties. Among them, the sulfates of copper, zinc, and manganese have good solubility, and sulfur is also required by plants, so sulfates are generally used. Boron is provided by borax, and molybdenum by sodium molybdate. The amount of chlorine required is very small, and the chlorine in the water source is basically sufficient.

[0053] When preparing nutrient solutions for hydroponics, different crops require different fertilizer conditions, and therefore the nutrient solution formulations also vary. Table 1 below lists several nutrient solution formulations, which can be selected according to needs or used as a reference.

[0054] Table 1 Formulas for various common plant nutrient solutions

[0055] The above formula specifies the nutrient ratio that the nutrient solution should contain when growing plants from seedlings using hydroponics. This nutrient ratio can also be used when raising seedlings, but the concentration of the nutrient solution should be appropriately reduced to prevent excessive salt content in the substrate from affecting the normal growth of seedlings, and excessive evaporation can also easily damage the leaves of seedlings.

[0056] In practice, the nutrient solution pool needs to be tested every day, and the dataset contains data on the complete plant growth cycle.

[0057] In the preferred embodiment, step S1 includes:

[0058] Plants are divided into multiple growth stages, including budding, growth, flowering, and fruiting.

[0059] Multiple preset time points are allowed. The nutrients in the nutrient solution are one or more of the following: nitrogen, phosphorus, potassium, calcium, iron, magnesium, sulfur, boron, zinc, copper, molybdenum, and chlorine. The nutrients exist in the form of salts. The proportion of nutrients in each growth stage is determined based on the nutrient requirements of the plant at each growth stage.

[0060] Sensors are used to detect the ion concentration data of nutrients in the nutrient solution.

[0061] The daily water consumption of plants was recorded and correlated with the ion concentration data of nutrients.

[0062] In this embodiment, step S1 sets the plant's growth milestones, such as the budding stage, growth stage, flowering stage, and fruiting stage. Within each growth milestone, such as 9:00 AM and 3:00 PM daily, sensors are used to detect the nutrient composition of the nutrient solution containing the plant, including the concentrations of nitrogen, phosphorus, and potassium, while simultaneously recording water consumption. Nutrient composition detection also includes the detection of specific elements. Microcirculation branches are established based on the specific elements required by the plant species, and the concentration of these specific components is monitored.

[0063] In this embodiment, the plant growth cycle is divided into different stages, and the corresponding ratios are determined based on the plant's nitrogen, phosphorus, and potassium requirements at different growth stages. Multiple sensors and instruments can be used for data acquisition to improve data reliability; for nutrient concentration detection, analytical instruments are combined to improve detection accuracy.

[0064] For example, the ion detection module uses sensors to detect the ion concentration of the hydroponic nutrient solution, defining this ion concentration as the initial concentration. Based on the hydroponic stage and plant species, combined with data from a local database, it analyzes the impact of the current plant species on the nutrient composition of the solution at the corresponding hydroponic stage. Then, the ion detection module continuously monitors the ion concentration of the hydroponic nutrient solution using sensors, defining this as the real-time ion concentration. The real-time ion concentration is compared with the standard ion concentration; if the comparison results are inconsistent, a reminder to adjust the nutrient solution is issued.

[0065] There are many types of sensors for monitoring fertilizer parameters such as nitrogen, phosphorus, and potassium. Commonly used specialized sensors include conductivity sensors, pH sensors, nitrogen-phosphorus-potassium ion selective electrode sensors, and temperature sensors. In this embodiment, the detection of common nitrogen, phosphorus, and potassium parameters can be achieved using corresponding sensors. Specific models of nitrogen, phosphorus, and potassium ion selective electrode sensors that can be used in this embodiment are as follows:

[0066] 1. Nitrogen ion selective electrode sensor: Thermo Scientific Orion 9512BNWP nitrogen ion selective electrode or Hach LZN7010.97 nitrogen ion selective electrode can be used.

[0067] 2. Phosphorus ion selective electrode sensor: Thermo Scientific Orion 9512HP phosphorus ion selective electrode can be used.

[0068] 3. Potassium ion selective electrode sensor: Thermo Scientific Orion 9512GP potassium ion selective electrode.

[0069] Common specialized sensors can be used to measure the concentration of corresponding ions, helping to monitor the content of elements such as nitrogen, phosphorus, and potassium in nutrient solutions, thereby adjusting the nutrient solution ratio and promoting plant growth and development. When selecting a sensor model, a suitable model can be chosen based on specific needs and laboratory conditions. Specialized sensors and detection instruments are also available in the current technology.

[0070] In the preferred embodiment, step S2 can also perform growth curve fitting, which involves linearly fitting the obtained water consumption and nutrient composition to obtain the growth curves of both over time.

[0071] Step S3 uses a neural network to build a detection model, uses a dataset of growth curves to train the model, and saves the optimal model corresponding to the growth curve.

[0072] The acquired water consumption and nutrient composition data are correlated and preprocessed, such as through data cleaning and noise reduction. Then, the data is divided into training and testing sets according to a preset ratio, for example, 8:2. Next, a linear fit is performed on the acquired water consumption and nutrient composition data using the least squares method. The nutrients are nitrogen, phosphorus, and potassium, allocated proportionally. Each set of nitrogen, phosphorus, and potassium data serves as a feature value, and each feature value is linearly fitted to its corresponding water consumption to obtain a growth curve for both over time. The model is then trained based on these growth curves to obtain the optimal model.

[0073] In the preferred embodiment, step S2 includes: linearly fitting the obtained water consumption and nutrient composition, using the least squares method for the fitting formula, wherein the nutrients are distributed proportionally, and some components are selected as characterization data, each group of nutrient composition characterization data is used as a feature value, and each feature value is linearly fitted with the corresponding water consumption; the characterization data are nitrogen, phosphorus, and potassium.

[0074] In this embodiment, nitrogen, phosphorus, and potassium are allocated in proportion, and each set of nitrogen, phosphorus, and potassium data is used as a feature value to monitor and adjust the nutritional status of plants in real time. Then, linear fitting is performed with water consumption to better understand the plant's needs for different nutrients and to carry out more effective plant growth management in order to obtain the optimal baseline growth curve.

[0075] In the preferred scheme, a neural network is used to establish a detection model to predict the relationship between the characterization data and the corresponding water consumption.

[0076] This embodiment employs a neural network with multiple input nodes (nitrogen, phosphorus, and potassium usage) and one output node (water consumption). It uses an attention-based MLP (Multi-Layer Fully Connected Network) model to analyze and predict based on the characteristics of the current data.

[0077] The specific steps for this implementation are as follows:

[0078] 1) Obtain the current water consumption and the concentrations of nitrogen, phosphorus, and potassium.

[0079] 2) Construct feature data as data = [moisture, (nitrogen, phosphorus, potassium)], and it is ordered in time.

[0080] 3) Input the MLP model to obtain the predicted value.

[0081] Furthermore, employing appropriate activation functions, such as linear activation functions, ensures that the relationships learned by the network are linear. In practical neural network models, non-linear activation functions (such as Sigmoid, ReLU, etc.) are often used to increase the model's expressive power, thereby enabling it to better capture complex relationships in the data.

[0082] Step S3: Build a detection model using a neural network, train the model using the growth curve dataset, and save the optimal model corresponding to the growth curve. Model training utilizes iterative training to train the detection model, including setting the number of iterations, inputting the training set into the detection model, the detection model making forward predictions, calculating the loss function on the prediction results, using the obtained loss value for backpropagation to optimize the detection model, and saving the optimal detection model when the current detection model outperforms the previous one.

[0083] In the preferred embodiment, the model training utilizes iterative training to train the detection model, including:

[0084] Obtain the dataset, divide it into training and test sets, and perform data preprocessing.

[0085] Set the number of iterations for training; input the preprocessed training set into the detection model.

[0086] The detection model makes forward predictions.

[0087] The loss function is calculated based on the prediction results.

[0088] The obtained loss value is used for backpropagation to optimize the detection model.

[0089] If the current detection model is better than the previous detection model, save the optimal detection model.

[0090] In this embodiment, the dataset forms the basis for training the model, including input features and corresponding labels. Data preprocessing is performed to clean, transform, and standardize the data to better suit the model. Dividing the dataset into training and test sets aims to verify the model's generalization ability during training.

[0091] In this embodiment, the number of iterations for training is set to 100.

[0092] Forward prediction refers to inputting data into a pre-trained model to obtain the model's prediction results. Loss functions are used to measure the difference between the model's predictions and the true labels; commonly used loss functions include mean squared error and cross-entropy.

[0093] During training, the model that performs best on the validation set is usually saved for subsequent testing and prediction on new data.

[0094] In practical applications, adjustments and optimizations can be made according to specific circumstances, such as adjusting the number of iterations or selecting appropriate optimization algorithms.

[0095] During the training of the detection model, the optimizer can be SGD, Adam, or AdamW, and the size of the input image and the parameter selection during the training process can be changed.

[0096] In the preferred scheme, adjusting the nutrient solution ratio according to the concentration predicted by the optimal model includes:

[0097] Based on the difference between the current nutrient concentration and the concentration predicted by the optimal model, the amount of nutrient solution and the amount of water output are controlled respectively. The total amount of nutrients required is calculated, and the nutrient solution is mixed with water to adjust the nutrient solution. The solution is replenished in multiple times and tested at fixed intervals. When the concentration predicted by the optimal model is reached, the adjustment is stopped and the water level in the plant planting pipeline is maintained.

[0098] In this embodiment, by adjusting the nutrient solution ratio based on the prediction results of the optimal model, it is possible to ensure that the plants receive the best nutrient supply during their growth, which helps to improve the plant's growth efficiency and yield.

[0099] Replenishing fertilizer multiple times and conducting regular testing can ensure the accuracy and stability of the fertilizer mix, while avoiding over- or under-fertilization.

[0100] Maintaining the water level in the plant planting pipeline ensures that the plant roots can fully absorb the nutrient solution, promoting healthy plant growth.

[0101] In the preferred method, the nutrient solution is replenished multiple times and monitored at fixed intervals. Specifically, this is done 4-6 times, with each fixed interval not less than the length of the plant planting pipeline divided by the current flow rate of the nutrient solution in the pot. The frequency of nutrient supply can be adjusted according to the actual growth of the plant to ensure even nutrient delivery.

[0102] In step S4, nutrient components are monitored in real time at different time points using the optimal model. When the difference between the nutrient concentration and the concentration predicted by the optimal model exceeds a set error threshold, nutrient solution is added. The nutrient solution ratio is adjusted according to the concentration predicted by the optimal model. This includes controlling the nutrient solution volume and water output based on the difference between the current nutrient concentration and the optimal model's predicted concentration, calculating the required total amount of nutrients, and mixing the nutrient solution with water to prepare the nutrient solution. Multiple additions are performed at fixed intervals, and monitoring is conducted. When the concentration predicted by the optimal model is reached, the preparation process ends, and the water level in the plant planting pipeline is maintained.

[0103] In the preferred embodiment, the power of the nutrient solution pump and the first water pump is adjusted, and after they are mixed in the mixing tank, the second water pump is started for irrigation.

[0104] First, start the nutrient solution pump and the first water pump to draw a certain volume of nutrient solution V1 and the added water volume V2 for mixing, and finally obtain the required nutrient solution concentration ρ, as shown in the formula:

[0105] In the formula, ρ1 is the high nutrient solution concentration, V1 is the high concentration nutrient solution, and V2 is the volume of water added.

[0106] The current flow rate of the nutrient solution is obtained by calculating the output of the second water pump.

[0107] Then, the second water pump is started to draw out the prepared nutrient solution for irrigation.

[0108] In the preferred embodiment, the nutrient component detection in step S1 also includes the detection of special elements. A microcirculation branch is established based on the special elements required by the plant species, and the concentration of the special components at the current time is detected.

[0109] By monitoring the concentration of specific components, we can promptly understand the supply of specific elements required by plants. If the test results indicate insufficient concentration of specific elements, targeted supplementation can be carried out through the microcirculation branch to ensure that plants receive the necessary nutrients. This better meets the specific element requirements of different plant species and helps improve plant growth quality and yield.

[0110] This embodiment provides a scientific and efficient nutrient management solution for hydroponic plants through a real-time, accurate detection and adjustment mechanism. It not only improves plant growth rate and quality but also reduces the frequency and difficulty of human intervention, providing strong support for the sustainable development of modern agriculture.

[0111] Example 2

[0112] Further illustrating with reference to Example 1, as shown in Figure 2, a mobile detection system for components of hydroponic plant nutrient solution includes:

[0113] The acquisition module is used to set multiple growth node time periods for the plant. Within the set time interval of each growth node time period, the nutrient composition of the plant's nutrient solution is detected, and the water consumption is recorded.

[0114] The data preprocessing module is used to correlate the acquired water consumption and nutrient composition in the dataset, perform data preprocessing, and then divide the dataset into training and test sets according to a preset ratio.

[0115] The model building module is used to build a detection model using a neural network, train the model using a training set, and save the optimal model.

[0116] The prediction module is used to detect nutrient components at different time points in real time. It uses the optimal model for detection. When the difference between the concentration of the nutrient component and the concentration predicted by the optimal model is greater than the set error threshold, nutrient solution is added and the nutrient solution ratio is adjusted according to the concentration predicted by the optimal model.

[0117] In this embodiment, the nutrient component detection module not only includes conventional nitrogen, phosphorus, and potassium concentrations, but also detects specific elements required by plant species to ensure comprehensive acquisition of plant nutritional needs information.

[0118] The data uses time-based growth curves, which can reflect the changes in the plant's water and nutrient requirements at different growth stages. The data preprocessing module processes the data accordingly to ensure its accuracy and completeness.

[0119] The model building module employs neural network technology, using a dataset of growth curves for model training. Through iterative training, the model's predictive ability is continuously optimized, ultimately saving an optimal model corresponding to the growth curve. This model can predict the plant's nutrient solution requirements in real time, providing a scientific basis for subsequent adjustments.

[0120] The prediction module is the core of the entire system. It monitors nutrient composition at different time points in real time and uses the optimal model to make predictions. When the difference between the actual detected nutrient concentration and the model-predicted concentration exceeds a set error threshold, the system triggers a nutrient solution replenishment and adjustment mechanism. This mechanism precisely controls the amount of nutrient solution replenished and the amount of water discharged based on the difference, ensuring that the composition of the nutrient solution is always kept within the optimal range.

[0121] The above embodiments are merely preferred technical solutions of the present invention and should not be considered as limitations on the present invention. The scope of protection of the present invention should be limited to the technical solutions described in the claims, including equivalent substitutions of the technical features described in the claims. That is, equivalent substitutions and improvements within this scope are also within the scope of protection of the present invention.

Claims

1. A method for detecting a component of a hydroponic plant nutrient solution, characterized by, The method comprises the following steps: S1: obtaining a data set, setting multiple growth node time periods of a plant, detecting the nutrient components of the nutrient solution of the plant in a set time interval of each growth node time period, and recording the water consumption amount; S2: corresponding the obtained water consumption amount and nutrient components in the data set, performing data preprocessing, and dividing into a training set and a test set according to a preset proportion; S3: establishing a detection model by using a neural network, training the model by using the training set, and saving an optimal model; S4: real-time detection of the nutrient components at different time points, detection by using the optimal model, supplement of the nutrient solution when the difference between the concentration of the nutrient components and the predicted concentration of the optimal model is greater than a set error threshold, and adjustment of the nutrient solution ratio according to the predicted concentration of the optimal model.

2. The method for detecting the components of a hydroponic plant nutrient solution according to claim 1, characterized by, The S2 can also perform growth curve fitting, linear fitting of the obtained water consumption amount and nutrient components, and obtaining the growth curves of the two with respect to time; The S3 establishes a detection model by using a neural network, uses a data set of the growth curves to train the model, and saves an optimal model corresponding to the growth curves.

3. The method of claim 1, wherein the component is selected from the group consisting of nitrogen, phosphorus, potassium, calcium, magnesium, iron, manganese, copper, zinc, boron, molybdenum, and chlorine. The S1 comprises: dividing the plant into multiple growth node time periods of a germination period, a growth period, a flowering period and a fruiting period; presetting multiple set time points, the nutrient components including nutrient elements of the nutrient solution being a combination of one or more of nitrogen, phosphorus, potassium, calcium, iron, magnesium, sulfur, boron, zinc, copper, molybdenum and chlorine, the nutrient components existing in the form of salt, and determining the proportion of the nutrient components in each growth stage according to the demand characteristics of the plant for the nutrient components in each growth node time period; detecting the ion concentration data of the nutrient components of the nutrient solution by using a sensor; wherein the water consumption amount of the plant is recorded per day and corresponds to the ion concentration data of the nutrient components.

4. The method for detecting the components of a hydroponic plant nutrient solution according to claim 3, characterized by, The S2 comprises: linear fitting of the obtained water consumption amount and nutrient components, a fitting formula, use of the least square method, proportional distribution of the nutrient components, selection of part of the components as representation data, use of each set of representation data of the nutrient components as a characteristic value, linear fitting of each characteristic value and the corresponding water consumption amount; and the representation data is nitrogen, phosphorus and potassium.

5. The method of claim 1, wherein the component is selected from the group consisting of nitrogen, phosphorus, potassium, calcium, magnesium, iron, manganese, copper, zinc, boron, molybdenum, and chlorine. The detection model is established by using a neural network to predict the relationship between the representation data and the corresponding water consumption amount.

6. The method of claim 5, wherein the component is selected from the group consisting of nitrogen, phosphorus, potassium, calcium, magnesium, iron, manganese, copper, zinc, boron, molybdenum, and chlorine. The model training trains the detection model by using iterative training, including: setting the number of iterations, inputting the training set into the detection model; the detection model performs forward prediction; loss function calculation is performed on the predicted result; the obtained loss value is used for back propagation to optimize the detection model; when the current detection model is better than the previous detection model, the optimal detection model is saved.

7. The method of claim 1, wherein the component is selected from the group consisting of nitrogen, phosphorus, potassium, calcium, magnesium, iron, manganese, copper, zinc, boron, molybdenum, and chlorine. The adjustment of the nutrient solution ratio according to the predicted concentration of the optimal model comprises: controlling the nutrient solution amount and the water amount respectively according to the difference between the current nutrient component concentration and the predicted concentration of the optimal model, calculating the required total amount of nutrient components, mixing the nutrient solution and water to adjust the nutrient solution, supplementing in multiple times and detecting at a fixed time interval, and ending the adjustment and keeping the water level in the plant planting pipeline when the predicted concentration of the optimal model is reached.

8. The method of claim 7, wherein the component is selected from the group consisting of nitrogen, phosphorus, potassium, calcium, magnesium, iron, manganese, copper, zinc, boron, molybdenum, and chlorine. The multiple times of replenishment and the detection with a fixed time interval are specifically: 4-6 times, and the fixed time interval is not less than the length of the plant planting pipeline divided by the flow rate of the current nutrient solution in the pot.

9. The method of claim 8, wherein the component is selected from the group consisting of nitrogen, phosphorus, potassium, calcium, magnesium, iron, manganese, copper, zinc, boron, molybdenum, and chlorine. The power of the nutrient solution pump and the first water pump is adjusted, and after mixing in the mixing tank, the second water pump is started for irrigation.

10. The method of claim 1, wherein the component is selected from the group consisting of nitrogen, phosphorus, potassium, calcium, magnesium, iron, manganese, copper, zinc, boron, molybdenum, and chlorine. The detection of the nutrient components in S1 also includes special element detection, and a microcirculation branch is established according to the special elements required by the plant species according to the concentration of the detected special components.

11. A mobile detection system for hydroponic plant nutrient solution components, comprising: a mobile detection system according to any one of claims 1 to 10; and a mobile detection system according to any one of claims 1 to 10. It comprises: An acquisition module is configured to set a plurality of growth node time periods of a plant, detect nutrient components of the nutrient solution of the plant in a set time interval of each growth node time period, and record water consumption; A data preprocessing module is configured to correspond the acquired water consumption and nutrient components in the data set, perform data preprocessing, and divide the data set into a training set and a test set according to a preset proportion; A model building module is configured to establish a detection model by using a neural network, train the model by using the training set, and save an optimal model; A prediction module is configured to detect nutrient components at different time points in real time, detect the nutrient components by using the optimal model, supplement the nutrient solution when a difference between a concentration of the nutrient components and a predicted concentration of the optimal model is greater than a set error threshold, and adjust a ratio of the nutrient solution according to the predicted concentration of the optimal model.

Images