Power demand response scheduling method based on dynamic prediction model

Abstract

A power demand response scheduling method based on a dynamic prediction model comprises the following steps: acquiring environmental data, crop growth demand data and energy supply data inside and outside a greenhouse in real time, and preprocessing the environmental data, the crop growth demand data and the energy supply data to construct a data set; and constructing a gas exchange model and a kinetic nutrition transfer model based on the data set, and outputting parameters and states of the gas exchange model and the kinetic nutrition transfer model. And analyzing parameters and states output based on the gas exchange model and the kinetic nutrition transmission model through a Bayesian network to predict the change trend of the environmental data in the greenhouse in the first time period in the future, and outputting a prediction result of the future environmental data. And formulating an energy scheduling scheme according to the prediction result of the future environmental data through a multi-objective optimization algorithm, and determining an energy use strategy of the heating, refrigeration and illumination systems so as to output an optimal energy scheduling scheme.

Description

Technical Field

[0001] The present invention belongs to the technical field of intelligent agricultural production and energy dispatching, and more specifically, relates to a power demand response dispatching method based on a dynamic prediction model. Background Art

[0002] As the global agricultural modernization process accelerates, agricultural greenhouses, as important facilities to ensure efficient and stable crop production, have an increasing demand for precise control of environmental conditions. Environmental parameters such as temperature, humidity and light in the greenhouse need to be maintained within a specific range to promote optimal crop growth. To achieve this goal, greenhouses rely extensively on heating, cooling and lighting systems, which have huge and highly volatile energy requirements and are more susceptible to external climate conditions. Therefore, how to efficiently and intelligently manage greenhouse energy use has become the key to improving agricultural production efficiency and reducing operating costs.

[0003] At present, the energy management of agricultural greenhouses mainly adopts rule-based control methods and traditional demand response (DR) systems. These systems usually rely on preset environmental parameters and energy use strategies, and lack the ability to dynamically respond to real-time data. With the development of information technology and intelligent control technology, model-based prediction-based power demand response systems have gradually been introduced into greenhouse energy management. Prediction models are used to predict future energy demand and environmental changes, and energy scheduling plans are optimized accordingly to achieve efficient energy utilization and minimize costs. At the same time, some greenhouses have begun to combine renewable energy (such as solar energy) with traditional energy (such as grid power supply) to improve the sustainability and stability of energy supply.

[0004] However, in the existing greenhouse energy management methods, due to the diversity and sensitivity of crop growth cycles, different crops and their different growth stages have different and highly sensitive requirements for environmental conditions, and the existing DR system lacks sufficient flexibility in adjusting energy use, making it difficult to accurately match the diverse crop needs and dynamically changing growth cycles. In addition, due to the difficulty of mixed management of renewable energy and traditional energy, agricultural greenhouses often need to use renewable energy such as solar energy and traditional energy such as grid power supply at the same time, and the existing DR system still has certain technical difficulties in efficiently coordinating and managing the mixed use of these two energy sources, which to a certain extent affects the energy utilization efficiency and the overall stability of the greenhouse system. Summary of the invention

[0005] In order to solve the deficiencies in the prior art, the purpose of the present invention is to solve the above-mentioned defects and further propose a power demand response scheduling method based on a dynamic prediction model.

[0006] The present invention adopts the following technical solution.

[0007] The first aspect of the present invention discloses a method for responding to power demand based on a dynamic prediction model, the method comprising: Acquire environmental data inside and outside the greenhouse, crop growth demand data, and energy supply data in real time, and pre-process the environmental data, crop growth demand data, and energy supply data to construct a data set; Constructing a gas exchange model and a kinetic nutrient transfer model based on the data set, and outputting parameters and states of the gas exchange model and the kinetic nutrient transfer model; Analyze the parameters and states output by the gas exchange model and the kinetic nutrient transfer model through a Bayesian network to predict the change trend of the environmental data in the greenhouse in the first time period in the future, and output the prediction results of the future environmental data; Formulate an energy scheduling plan based on the prediction results of the future environmental data through a multi-objective optimization algorithm, determine the energy use strategy of the heating, cooling and lighting systems, and output the optimal energy scheduling plan; Among them, the environmental data includes temperature, humidity, light intensity, carbon dioxide concentration and nutrient solution flow rate, the energy supply data includes solar energy and electrical energy, the preprocessing includes missing value processing, outlier detection and data smoothing processing, the gas exchange model is used to simulate the dynamic changes of carbon dioxide concentration and oxygen in the greenhouse, and the kinetic nutrient transfer model is used to simulate the transmission and distribution of nutrient solution in the hydroponic system.

[0008] Furthermore, the real-time acquisition of environmental data inside and outside the greenhouse, crop growth demand data, and energy supply data, and preprocessing of the environmental data, crop growth demand data, and energy supply data to construct a data set includes: Monitoring the environmental data, crop growth demand data and energy supply data inside and outside the greenhouse in real time through multiple sensors and energy monitoring equipment, and constructing an original data set based on the environmental data, crop growth demand data and energy supply data; Filling missing values ​​in the original data set with the mean of the corresponding data column in the original data set as the filling value, and detecting outliers in the original data set using the three sigma principle, so that when an outlier is detected, the outlier is replaced with the median of the corresponding column in the original data set; The raw data set is smoothed by a moving average method, and the cleaned environmental data, crop growth demand data, and energy supply data are integrated to output an integrated data set.

[0009] Furthermore, the real-time acquisition of environmental data inside and outside the greenhouse, crop growth demand data, and energy supply data, and preprocessing the environmental data, crop growth demand data, and energy supply data to construct a data set also includes: Standardizing and normalizing the integrated data set to eliminate the influence of different dimensions and magnitudes in the integrated data set on the gas exchange model and the kinetic nutrient transfer model, and outputting the converted data set; The standardized processing expression is: The normalized expression is: In the formula, and are the i-th row and j-th list elements respectively The standardized and normalized results of is the element in the i-th row and j-th column of the integrated data set, is the mean of the jth column, is the standard deviation of the jth column, and are the minimum and maximum values ​​of the jth column respectively.

[0010] Furthermore, the gas exchange model and the kinetic nutrient transfer model are constructed based on the data set, and the parameters and states of the gas exchange model and the kinetic nutrient transfer model are output, including: Acquire the first basic parameters required by the gas exchange model from the converted data set, the first basic parameters including the carbon dioxide concentration, oxygen concentration, ventilation rate and photosynthesis rate inside and outside the greenhouse, and construct the first basic data set of the gas exchange model; Based on the first basic data set, a kinetic gas exchange equation is established to describe the change of carbon dioxide concentration and oxygen concentration in the greenhouse over time. The expression of the kinetic gas exchange equation is: In the formula, is the rate of change of carbon dioxide concentration over time, is the rate of change of oxygen concentration over time, is the total volume of greenhouse gases, is the ventilation rate, is the photosynthesis rate, is the concentration of carbon dioxide entering the greenhouse, is the oxygen concentration entering the greenhouse, and are the carbon dioxide concentration and oxygen concentration in the greenhouse at time t, is the total photosynthetic area of ​​the crop at time t.

[0011] Furthermore, the step of constructing a gas exchange model and a kinetic nutrient transfer model based on the data set, and outputting parameters and states of the gas exchange model and the kinetic nutrient transfer model, further includes: Acquire the second basic parameters required by the kinetic nutrient transfer model from the converted data set, the second basic parameters including nutrient solution flow rate, nutrient concentration in the nutrient solution, total plant root area and nutrient absorption rate, and construct a second basic parameter set; Based on the second basic parameter set, a kinetic nutrient transfer equation describing the nutrient changes in the nutrient solution is established. The expression of the kinetic nutrient transfer equation is: In the formula, is the rate of change of nutrient concentration over time, is the volume of the nutrient solution storage tank, is the nutrient concentration of the input nutrient solution, is the total root area of ​​the plant, is the nutrient solution flow rate, is the nutrient concentration in the nutrient solution at time t, is the nutrient absorption rate; The gas exchange model and the kinetic nutrient transfer model are calibrated by minimizing the error between the model prediction value and the actual monitoring value, and the first basic parameter set and the second basic parameter set are divided into a training set and a validation set by cross-validation or holdout method, and the determination coefficient and the mean square error index are calculated to verify the gas exchange model and the kinetic nutrient transfer model, and the calibrated model output parameters and status are output.

[0012] Furthermore, the Bayesian network is used to analyze the parameters and states outputted by the gas exchange model and the kinetic nutrient transfer model to predict the changing trend of the environmental data in the greenhouse in the first time period in the future, and output the prediction results of the future environmental data, including: Constructing a Bayesian network training data set based on the calibrated model output parameters and states, and determining the dependency between the environmental data in the Bayesian network according to the Bayesian network training data set through a structural learning algorithm to construct a Bayesian network structure; The conditional probability table of each node is calculated according to the Bayesian network training data set using a maximum likelihood estimation or Bayesian estimation algorithm, and the Bayesian network is trained in combination with the Bayesian network structure and the Bayesian network training data set, and the model structure and parameters are adjusted through cross-validation to obtain a trained Bayesian network model; Among them, the Bayesian network structure consists of a node set and an edge set, the node set represents each environmental data, the edge set represents the directed dependency relationship between nodes, the structural learning algorithm includes a scoring search algorithm and a constraint-based algorithm, and the trained Bayesian network model is used to predict and output the prediction results of the future environmental data based on current or historical environmental data.

[0013] Furthermore, the multi-objective optimization algorithm includes an optimization objective function and a predefined set of constraints, wherein the optimization objective function includes a plurality of sub-objectives, wherein the sub-objectives include minimizing energy costs, maximizing energy utilization efficiency, and ensuring environmental stability; The energy scheduling scheme is formulated according to the prediction results of the future environmental data by the multi-objective optimization algorithm, and the energy use strategy of the heating, cooling and lighting systems is determined to output the optimal energy scheduling scheme, including: Based on the weight coefficient of each sub-objective in the optimization objective function, the sub-objectives are weighted and summed to obtain a comprehensive optimization objective function, and the configuration parameters are determined by a non-dominated sorting genetic algorithm in combination with the constraint condition set to construct a configured multi-objective optimization algorithm; Calling the configured multi-objective optimization algorithm to generate a solution set of approximate Pareto optimal solutions based on the comprehensive optimization objective function and the constraint condition set, and screening out the optimal energy scheduling solution from the solution set according to actual needs and priorities; The configuration parameters include population size, number of iterations, crossover rate and mutation rate, and each solution in the solution set corresponds to an energy scheduling scheme.

[0014] The second aspect of the present invention discloses a power demand response system scheduling device based on model prediction, the device comprising: A data preprocessing module, used to obtain environmental data inside and outside the greenhouse, crop growth demand data, and energy supply data in real time, and preprocess the environmental data, crop growth demand data, and energy supply data to construct a data set; A model building module, used to build a gas exchange model and a kinetic nutrient transfer model based on the data set, and output parameters and states of the gas exchange model and the kinetic nutrient transfer model; A future environment prediction module is used to analyze the parameters and states output by the gas exchange model and the kinetic nutrient transfer model through a Bayesian network to predict the change trend of the environmental data in the greenhouse within the first time period in the future and output the prediction results of the future environmental data; An energy scheduling screening module is used to formulate an energy scheduling plan according to the prediction results of the future environmental data through a multi-objective optimization algorithm, determine the energy use strategy of the heating, cooling and lighting systems, and output the optimal energy scheduling plan; Among them, the environmental data includes temperature, humidity, light intensity, carbon dioxide concentration and nutrient solution flow rate, the energy supply data includes solar energy and electrical energy, the preprocessing includes missing value processing, outlier detection and data smoothing processing, the gas exchange model is used to simulate the dynamic changes of carbon dioxide concentration and oxygen in the greenhouse, and the kinetic nutrient transfer model is used to simulate the transmission and distribution of nutrient solution in the hydroponic system.

[0015] The third aspect of the present invention discloses a terminal, including a processor and a storage medium; the characteristics are: The storage medium is used to store instructions; The processor is used to operate according to the instructions to execute the steps of the method described in the first aspect.

[0016] The fourth aspect of the present invention discloses a computer-readable storage medium having a computer program stored thereon, characterized in that when the program is executed by a processor, the steps of the method described in the first aspect are implemented.

[0017] The beneficial effects of the present invention are that, compared with the prior art, the present invention has the following advantages: By constructing a gas exchange model and a kinetic nutrient transfer model, and combining it with a Bayesian network to predict environmental data for a period of time in the future, and formulating and screening the optimal energy scheduling plan based on the prediction results through a multi-objective optimization algorithm, energy can be flexibly scheduled in greenhouses where the diversity and sensitivity of crop growth cycles are high, thereby improving the accuracy of matching energy with the growth needs of diversified crops, and at the same time improving energy utilization efficiency and the stability of the energy scheduling system in scenarios where renewable energy and traditional energy are used together. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a flow chart of the power demand response scheduling method based on the dynamic prediction model provided by the present invention. DETAILED DESCRIPTION

[0019] The present application is further described below in conjunction with the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and cannot be used to limit the protection scope of the present application.

[0020] like Figure 1 As shown, in one embodiment, a power demand response scheduling method based on a dynamic prediction model includes the following steps: Step S110, acquiring environmental data inside and outside the greenhouse, crop growth demand data, and energy supply data in real time, and preprocessing the environmental data, crop growth demand data, and energy supply data to construct a data set.

[0021] Among them, environmental data include temperature, humidity, light intensity, carbon dioxide concentration and nutrient solution flow, energy supply data include solar energy and electrical energy, and preprocessing includes missing value processing, outlier detection and data smoothing.

[0022] In some embodiments, the power demand response scheduling method based on the dynamic prediction model provided by the present invention obtains environmental data inside and outside the greenhouse, crop growth demand data, and energy supply data in real time, and pre-processes the environmental data, crop growth demand data, and energy supply data to construct a data set, specifically including the following steps: Step S111, monitor the environmental data, crop growth demand data and energy supply data inside and outside the greenhouse in real time through multiple sensors and energy monitoring equipment, and construct an original data set based on the environmental data, crop growth demand data and energy supply data.

[0023] Step S112, using the mean of the corresponding data column in the original data set as the filling value to fill the missing values ​​in the original data set, and using the three sigma principle to detect outliers in the original data set, so that when an outlier is detected, the outlier is replaced with the median of the corresponding column in the original data set.

[0024] Step S113, smoothing the original data set by moving average method, and integrating the cleaned environmental data, crop growth demand data and energy supply data to output an integrated data set.

[0025] In some embodiments, the power demand response scheduling method based on the dynamic prediction model provided by the present invention obtains environmental data inside and outside the greenhouse, crop growth demand data, and energy supply data in real time, and pre-processes the environmental data, crop growth demand data, and energy supply data to construct a data set, and specifically includes the following steps: Step S114, standardize and normalize the integrated data set to eliminate the influence of different dimensions and magnitudes in the integrated data set on the gas exchange model and the kinetic nutrient transfer model, and output the converted data set.

[0026] The standardized processing expression is: The normalized expression is: In the formula, and are the i-th row and j-th list elements respectively The standardized and normalized results of is the element in the i-th row and j-th column of the integrated data set, is the mean of the jth column, is the standard deviation of the jth column, and are the minimum and maximum values ​​of the jth column respectively.

[0027] Step S120, constructing a gas exchange model and a kinetic nutrient transfer model based on the data set, and outputting parameters and states of the gas exchange model and the kinetic nutrient transfer model.

[0028] Among them, the gas exchange model is used to simulate the dynamic changes of carbon dioxide concentration and oxygen in the greenhouse, and the kinetic nutrient transfer model is used to simulate the transmission and distribution of nutrient solution in the hydroponic system.

[0029] In some embodiments, the power demand response scheduling method based on the dynamic prediction model provided by the present invention builds a gas exchange model and a kinetic nutrient transfer model based on a data set, and outputs parameters and states of the gas exchange model and the kinetic nutrient transfer model, specifically comprising the following steps: Step S121, obtaining the first basic parameters required by the gas exchange model from the converted data set, the first basic parameters including the carbon dioxide concentration, oxygen concentration, ventilation rate and photosynthesis rate inside and outside the greenhouse, and constructing the first basic data set of the gas exchange model.

[0030] Step S122, based on the first basic data set, a kinetic gas exchange equation is established to describe the change of carbon dioxide concentration and oxygen concentration in the greenhouse over time. The expression of the kinetic gas exchange equation is: In the formula, is the rate of change of carbon dioxide concentration over time, is the rate of change of oxygen concentration over time, is the total volume of greenhouse gases, is the ventilation rate, is the photosynthesis rate, is the concentration of carbon dioxide entering the greenhouse, is the oxygen concentration entering the greenhouse, and are the carbon dioxide concentration and oxygen concentration in the greenhouse at time t, is the total photosynthetic area of ​​the crop at time t.

[0031] In some embodiments, the power demand response scheduling method based on the dynamic prediction model provided by the present invention builds a gas exchange model and a kinetic nutrient transfer model based on a data set, and outputs parameters and states of the gas exchange model and the kinetic nutrient transfer model, and specifically includes the following steps: Step S123, obtaining the second basic parameters required by the dynamic nutrient transfer model from the converted data set, the second basic parameters include nutrient solution flow rate, nutrient concentration in the nutrient solution, total plant root area and nutrient absorption rate, and constructing the second basic parameter set Step S124, based on the second basic parameter set, a kinetic nutrient transfer equation is established to describe the nutrient changes in the nutrient solution. The expression of the kinetic nutrient transfer equation is: In the formula, is the rate of change of nutrient concentration over time, is the volume of the nutrient solution storage tank, is the nutrient concentration of the input nutrient solution, is the total root area of ​​the plant, is the nutrient solution flow rate, is the nutrient concentration in the nutrient solution at time t, is the nutrient absorption rate.

[0032] Step S125, calibrate the parameters of the gas exchange model and the kinetic nutrient transfer model by minimizing the error between the model prediction value and the actual monitoring value, and use cross-validation or holdout method to divide the first basic parameter set and the second basic parameter set into a training set and a validation set, calculate the determination coefficient and the mean square error index to verify the gas exchange model and the kinetic nutrient transfer model, and output the calibrated model output parameters and status.

[0033] Step S130, analyzing the parameters and states output by the gas exchange model and the kinetic nutrient transfer model through the Bayesian network to predict the changing trend of the environmental data in the greenhouse in the first time period in the future, and outputting the prediction results of the future environmental data.

[0034] In some embodiments, the power demand response scheduling method based on the dynamic prediction model provided by the present invention analyzes the parameters and states output by the gas exchange model and the kinetic nutrient transfer model through the Bayesian network to predict the change trend of the environmental data in the greenhouse in the first time period in the future, and outputs the prediction results of the future environmental data, which specifically includes the following steps: Step S131, constructing a Bayesian network training data set based on the calibrated model output parameters and states, and determining the dependency between various environmental data in the Bayesian network according to the Bayesian network training data set through a structural learning algorithm to construct a Bayesian network structure.

[0035] Step S132, using maximum likelihood estimation or Bayesian estimation algorithm to calculate the conditional probability table of each node according to the Bayesian network training data set, and training the Bayesian network in combination with the Bayesian network structure and the Bayesian network training data set, adjusting the model structure and parameters through cross-validation, and obtaining a trained Bayesian network model.

[0036] Among them, the Bayesian network structure consists of a node set and an edge set. The node set represents each environmental data, and the edge set represents the directed dependency relationship between nodes. The structural learning algorithm includes a scoring search algorithm and a constraint-based algorithm. The trained Bayesian network model is used to predict and output the prediction results of future environmental data based on current or historical environmental data.

[0037] Step S140, formulating an energy scheduling plan based on the prediction results of future environmental data through a multi-objective optimization algorithm, determining the energy use strategy of the heating, cooling and lighting systems, and outputting the optimal energy scheduling plan.

[0038] Among them, the multi-objective optimization algorithm includes an optimization objective function and a predefined set of constraints. The optimization objective function contains multiple sub-objectives, including minimizing energy costs, maximizing energy utilization efficiency, and ensuring environmental stability.

[0039] In some embodiments, the power demand response scheduling method based on the dynamic prediction model provided by the present invention formulates an energy scheduling plan according to the prediction results of future environmental data through a multi-objective optimization algorithm, determines the energy use strategy of the heating, cooling and lighting systems, and outputs the optimal energy scheduling plan, which specifically includes the following steps: Step S141, based on the weight coefficient of each sub-goal in the optimization objective function, weighted summation is performed on the sub-goals to obtain a comprehensive optimization objective function, and combined with the constraint set, the configuration parameters are determined through a non-dominated sorting genetic algorithm to construct a configured multi-objective optimization algorithm.

[0040] Step S142, calling the configured multi-objective optimization algorithm to generate a solution set that approximates the Pareto optimal solution based on the comprehensive optimization objective function and the constraint condition set, and screening out the optimal energy scheduling solution from the solution set according to actual needs and priorities.

[0041] The configuration parameters include population size, number of iterations, crossover rate, and mutation rate. Each solution in the solution set corresponds to an energy scheduling plan.

[0042] The above-mentioned electricity demand response scheduling method based on the dynamic prediction model constructs a gas exchange model and a kinetic nutrient transfer model, and combines the Bayesian network to predict the environmental data for a period of time in the future, and formulates and screens the optimal energy scheduling plan through a multi-objective optimization algorithm based on the prediction results. It can flexibly schedule energy in greenhouses with high diversity and sensitivity of crop growth cycles, thereby improving the accuracy of matching energy with diversified crop growth needs, and at the same time improving energy utilization efficiency and the stability of the energy scheduling system in scenarios where renewable energy and traditional energy are used together.

[0043] In a specific embodiment, the power demand response scheduling method based on the dynamic prediction model provided by the present invention includes steps 1 to 4: Step 1: Data collection and preprocessing, outputting preprocessed high-quality data sets for subsequent modeling and prediction.

[0044] Collect real-time environmental data inside and outside the greenhouse, including temperature, humidity, light intensity, carbon dioxide concentration, nutrient solution flow, etc. In addition, collect data on the physiological needs of different crops and their various growth stages, as well as the supply of renewable energy (such as solar energy) and traditional energy (such as the power grid).

[0045] Specifically, it includes steps 1.1 to 1.4: Step 1.1, real-time environmental data collection.

[0046] Real-time environmental data inside and outside the agricultural greenhouse are collected, including temperature, humidity, light intensity, carbon dioxide concentration, etc. In addition, crop growth demand data and supply data of renewable energy (such as solar energy) and traditional energy (such as power grid) are obtained. These data are monitored and recorded in real time through sensors and energy monitoring equipment installed at key locations in the greenhouse. Subsequently, the original data set is constructed based on the collected data. The original data set is represented as a matrix, in which each row is the temperature, humidity, light intensity, carbon dioxide concentration, renewable energy supply and traditional energy supply at the same time point, and each column is the temperature or humidity or light intensity or carbon dioxide concentration or renewable energy supply or traditional energy supply at different time points.

[0047] Step 1.2, data cleaning.

[0048] The original data set obtained in step 1.1 is cleaned to remove missing values ​​and outliers, and is smoothed to ensure the accuracy and consistency of the data.

[0049] In the process of missing value processing, for each element in the original data set matrix, if the corresponding element is missing, the mean of the column is used as the filling value. In the detection and processing of outliers, the three sigma principle is used to detect outliers. If an element is marked as an outlier, the median of the column is used to replace the outlier. In the process of data smoothing, the moving average method is used to smooth the data, and finally the cleaned data set is obtained.

[0050] Step 1.3, data integration.

[0051] The cleaned environmental data, crop growth demand data, and energy supply data obtained in step 1.2 are integrated to form a unified data set, namely, an integrated data set, to facilitate modeling for subsequent analysis.

[0052] Step 1.4, data conversion.

[0053] The integrated data set is standardized and normalized to obtain a converted data set, which is used to eliminate the impact of different dimensions and magnitudes on subsequent model building and improve data consistency and model convergence speed.

[0054] Specifically, the standardized processing expression is: The normalized expression is: In the formula, and are the i-th row and j-th list elements respectively The standardized and normalized results of is the element in the i-th row and j-th column of the integrated data set, is the mean of the jth column, is the standard deviation of the jth column, and are the minimum and maximum values ​​of the jth column respectively.

[0055] Step 2: Establish a gas exchange and kinetic nutrient transfer model, and output the parameters and states of the gas exchange model and the kinetic nutrient transfer model.

[0056] Based on the collected data, a gas exchange model and a kinetic nutrient transfer model are constructed. The gas exchange model is used to simulate the dynamic changes of carbon dioxide and oxygen in the greenhouse to ensure the gas supply required for crop photosynthesis. The kinetic nutrient transfer model simulates the transmission and distribution of nutrient solution in the hydroponic system to ensure that crops obtain balanced nutrition.

[0057] Specifically, it includes steps 2.1 to 2.5: Step 2.1, determine the basic parameters of the gas exchange model.

[0058] Based on the preprocessed high-quality data set, the basic parameters required for the gas exchange model are first determined. The basic parameters include the carbon dioxide concentration, oxygen concentration, ventilation rate and photosynthesis rate inside and outside the greenhouse. Then, the basic parameter set of the gas exchange model is constructed. The parameter set is represented as a vector composed of the carbon dioxide concentration entering the greenhouse, the oxygen concentration entering the greenhouse, the ventilation rate and the photosynthesis rate.

[0059] Step 2.2, try to solve the gas exchange kinetic equation.

[0060] Using the basic parameter set of gas exchange constructed in step 2.1, a kinetic equation is established to describe the temporal changes in carbon dioxide and oxygen concentrations in the greenhouse. This equation takes into account the gases brought in by ventilation, the carbon dioxide consumed by crop photosynthesis, and the oxygen released.

[0061] Specifically, the expression of the kinetic gas exchange equation is: In the formula, is the rate of change of carbon dioxide concentration over time, is the rate of change of oxygen concentration over time, is the total volume of greenhouse gases, is the ventilation rate, is the photosynthesis rate, is the concentration of carbon dioxide entering the greenhouse, is the oxygen concentration entering the greenhouse, and are the carbon dioxide concentration and oxygen concentration in the greenhouse at time t, is the total photosynthetic area of ​​the crop at time t.

[0062] Step 2.3, determine the basic parameters of the kinetic nutrient transport model.

[0063] Based on the preprocessed high-quality data set, the basic parameters required for the kinetic nutrient transfer model are determined, including nutrient solution flow rate, nutrient concentration, and plant absorption rate. Subsequently, the basic parameter set of the kinetic nutrient transfer model is constructed based on the basic parameters, which is represented as a vector composed of nutrient solution flow rate, nutrient concentration in nutrient solution, total plant root area, and nutrient absorption rate.

[0064] Step 2.4, establish the kinetic nutrient transport equation.

[0065] Using the basic parameter set of kinetic nutrient transport constructed in step 2.3, a kinetic equation is established to describe the temporal variation of nutrient concentration in the nutrient solution. This equation takes into account the flow of the nutrient solution, the absorption of nutrients by plants, and the recycling of nutrients.

[0066] Specifically, the expression of the kinetic nutrient transfer equation is: In the formula, is the rate of change of nutrient concentration over time, is the volume of the nutrient solution storage tank, is the nutrient concentration of the input nutrient solution, is the total root area of ​​the plant, is the nutrient solution flow rate, is the nutrient concentration in the nutrient solution at time t, is the nutrient absorption rate.

[0067] Step 2.5, parameter calibration and model verification.

[0068] The parameters of the gas exchange model and the kinetic nutrient transport model were calibrated using experimental and historical data to ensure the accuracy and reliability of the models. Statistical methods and error analysis were used to assess the degree of agreement between the model predictions and the actual data.

[0069] Parameter calibration is performed by minimizing the error between the model prediction value and the actual observation value, and its expression is: In the formula, is the model parameter vector, is the actual observation value at the i-th time point, is the model prediction value at the i-th time point.

[0070] In the process of model validation, the data set is divided into training set and validation set by cross-validation or holdout method, the prediction performance of the model is evaluated, and the coefficient of determination is calculated. and mean square error The indicator is expressed as: In the formula, is the average value of the actual observations, n is the total number of elements in the data set, and other variables are the same as above and will not be repeated here. Value and lower The value indicates that the model has good predictive performance.

[0071] Based on the above process, the calibrated parameters and states of the gas exchange model and the kinetic nutrient transfer model are finally output.

[0072] Step 3: Construct a Bayesian network to predict environmental parameters and output the probability prediction results of future environmental parameters.

[0073] The Bayesian network is used to model the dependency between environmental parameters during crop growth and make probabilistic predictions. The outputs of the gas exchange model and the kinetic nutrient transfer model are analyzed to predict the possible change trends of various environmental parameters in the greenhouse in the future.

[0074] Specifically, it includes steps 3.1 to 3.5: Step 3.1, prepare the training data for the Bayesian network.

[0075] Organize and prepare the training data required for the Bayesian network using the parameters and states of the gas exchange model and kinetic nutrient transfer model established in step 2. Ensure that all relevant environmental parameters and model state data are correctly extracted and formatted for use in Bayesian network modeling and training.

[0076] In this embodiment, the Bayesian network training data set can be represented as a matrix, each row of the matrix is ​​the carbon dioxide concentration, oxygen concentration, nutrient concentration, temperature, humidity and light intensity at the same time point, and each column is the carbon dioxide concentration or oxygen concentration or nutrient concentration or temperature or humidity or light intensity at a different time point.

[0077] Step 3.2, determine the structure of the Bayesian network.

[0078] Through the structural learning algorithm, the dependencies between the environmental parameters in the Bayesian network are determined. The appropriate structural learning method (such as a scoring-based search algorithm, a constraint-based algorithm, or a hybrid method) is selected to construct a network structure that reflects the mutual influence of variables in the actual greenhouse environment. The Bayesian network structure consists of a node set and an edge set. The node set represents each environmental data, and the edge set represents the directed dependency between nodes.

[0079] In this embodiment, the constraint-based algorithm and the scoring-based search algorithm are used in greenhouse management. The carbon dioxide concentration may affect the oxygen concentration and nutrient concentration, and the temperature may affect the humidity and light intensity. These dependencies are determined through structural learning to construct a network structure that reflects the mutual influence of variables in the actual environment.

[0080] Step 3.3, parameter learning and probability estimation.

[0081] After determining the Bayesian network structure, parameter learning is performed to estimate the conditional probability distribution of each node. The maximum likelihood estimation (MLE) or Bayesian estimation method is used to calculate the conditional probability table (CPT) of each node based on the training data DBN.

[0082] In this embodiment, the conditional probability table is used to represent the probability distribution of each node when its parent node takes a specific value, and its expression is: In the formula, is node i, is the parent node of node i, is the value of node i, It is the value combination of the parent node.

[0083] In greenhouse management, MLE or Bayesian estimation is used to calculate the probability distribution of carbon dioxide concentration, oxygen concentration, and nutrient concentration under different environmental conditions (such as different temperatures and humidity). and humidity When the carbon dioxide concentration The probability is 400ppm.

[0084] In this embodiment, the MLE method estimates conditional probability by direct counting, which is applicable to large-scale data sets and ensures that the parameters obtained by the Bayesian network reflect the actual environmental data. Bayesian estimation avoids the problem of probability being 0 by introducing prior information, and is applicable to situations with less or different data, which can improve the robustness of the model.

[0085] Step 3.4, model training and optimization.

[0086] Use the Bayesian network training data set to train the Bayesian network, optimize the network parameters, adjust the model structure and parameters through cross-validation or other validation methods, and improve the accuracy and robustness of the prediction.

[0087] In the cross-validation process, the dataset is divided into a training set and a validation set, and the training and validation process is repeated to select the model parameters with the best performance. In addition, hyperparameter tuning can adjust the hyperparameters in the Bayesian network (such as smoothing parameters) to optimize model performance.

[0088] In greenhouse management, cross-validation and hyperparameter adjustment are used to ensure that the Bayesian network model can accurately predict the changing trends of carbon dioxide, oxygen and nutrient concentrations under different light, temperature and humidity conditions. For example, by adjusting the smoothing parameters, the prediction accuracy of carbon dioxide under high light intensity is optimized. In addition, by minimizing the prediction error, it can ensure that the Bayesian network model accurately reflects the dependencies between environmental parameters. Cross-validation can improve the generalization ability of the model and avoid overfitting through multiple training and verification.

[0089] Step 3.5, probabilistic prediction of future environmental parameters.

[0090] The trained and optimized Bayesian network model is used to make probability predictions of various environmental data in the greenhouse in the future. That is, the current or historical environmental parameter status is input, the probability distribution of future parameters is calculated through Bayesian reasoning, and the probability prediction results of future environmental parameters are output.

[0091] Specifically, the Bayesian inference formula is: In the formula, is the environmental parameter at a future time point, is the current or historical environmental parameter state, m=6 is the number of future environmental parameters, For each future parameter The parent node of , the dependency relationship in the Bayesian network.

[0092] In greenhouse management, Bayesian reasoning is used to predict the probability distribution of these parameters in the future by using the current state of environmental parameters (such as current temperature, humidity, light intensity, etc.). Bayesian reasoning uses the known current state, network structure and conditional probability table to calculate the joint probability distribution of future parameters. The adjustment rate of each future parameter is determined by the state of its parent node, ensuring that the prediction results reflect the actual dependencies and dynamic changes.

[0093] Step 4: The multi-objective optimization algorithm formulates an energy scheduling plan and outputs the optimal energy scheduling plan.

[0094] Based on the prediction results of the Bayesian network, a multi-objective optimization algorithm is applied to balance multiple objectives (such as energy cost, crop yield, and environmental stability) to optimize the energy scheduling plan. The optimization algorithm will determine the energy use strategy for heating, cooling, and lighting systems and reasonably allocate the proportion of renewable energy and traditional energy.

[0095] Specifically, it includes steps 4.1 to 4.4: Step 4.1, define the optimization objectives and constraints.

[0096] The optimization objectives and constraints of energy scheduling are clarified. The main objectives include minimizing energy costs, maximizing energy efficiency, and ensuring environmental stability. Constraints include the requirements of greenhouse environmental parameters, the supply limits of renewable energy and traditional energy, and the physical limitations of equipment operation.

[0097] In this embodiment, the optimization objective function includes multiple sub-objectives, namely, minimizing energy cost, maximizing energy efficiency, and ensuring environmental stability. The sub-objective expressions are: In the formula, , , They are minimizing energy costs, maximizing energy efficiency and ensuring environmental stability. is the total number of scheduling time periods, is the unit cost of renewable energy at time t, is the amount of renewable energy used at time t, is the unit cost of traditional energy at time t, is the amount of traditional energy used at time t, is the total energy actually used at time t, is the total energy available at time t, is the degree of satisfaction of the actual environmental parameters in the greenhouse at time t, is the degree of satisfaction of the expected environmental parameters in the greenhouse at time t.

[0098] Step 4.2, construct a multi-objective optimization model.

[0099] Based on the defined optimization objectives and constraints, a multi-objective optimization model is constructed, and the weighted summation method is used to merge multiple objective functions into a single objective function so that the optimization algorithm can be applied to solve it.

[0100] Specifically, based on the weight coefficient of each sub-goal in the optimization objective function, the sub-goals are weighted and summed to obtain a comprehensive optimization objective function. Combined with the constraint set, the configuration parameters are determined through a non-dominated sorting genetic algorithm to construct a configured multi-objective optimization algorithm.

[0101] Step 4.3, select a multi-objective optimization algorithm suitable for energy scheduling, such as particle swarm optimization, genetic algorithm or non-dominated sorting genetic algorithm. First, configure the algorithm parameters, such as population size, number of iterations, crossover rate and mutation rate, to ensure the efficiency and effect of the optimization process, and obtain the configured multi-objective optimization algorithm model.

[0102] Step 4.4, perform multi-objective optimization and generate a solution set.

[0103] Run the configured multi-objective optimization algorithm to generate a set of approximate Pareto optimal solutions to output the corresponding solution set, which achieves the best balance between different objectives for further selection and application. Each solution in the solution set is an energy scheduling scheme that defines the energy use strategy at each time point. In addition, the solution set represents different trade-off schemes, and users can choose the most appropriate scheme according to actual needs.

[0104] Step 4.5, screen and determine the optimal energy scheduling plan.

[0105] The optimal energy scheduling scheme is selected from the generated solution set, and the scheme that best meets the target balance is selected based on actual needs and priorities, and ensures that it meets all constraints.

[0106] Among them, the optimal energy scheduling solution It can be expressed as: In the formula, is the optimal amount of renewable energy used at time t, is the optimal amount of traditional energy used at time t, and other variables are the same as above and will not be repeated here.

[0107] The following describes the power demand response system scheduling device based on model prediction provided by the present invention. The power demand response system scheduling device based on model prediction described below and the power demand response scheduling method based on dynamic prediction model described above can refer to each other.

[0108] In one embodiment, a power demand response system scheduling device based on model prediction includes a data preprocessing module, a model building module, a future environment prediction module, and an energy scheduling screening module.

[0109] The data preprocessing module is used to obtain the environmental data inside and outside the greenhouse, the crop growth demand data and the energy supply data in real time, and preprocess the environmental data, the crop growth demand data and the energy supply data to construct a data set.

[0110] The model building module is used to build a gas exchange model and a kinetic nutrient transfer model based on the data set, and output the parameters and states of the gas exchange model and the kinetic nutrient transfer model.

[0111] The future environment prediction module is used to analyze the parameters and states output by the gas exchange model and the kinetic nutrient transfer model through the Bayesian network to predict the changing trend of the environmental data in the greenhouse in the first time period in the future and output the prediction results of the future environmental data.

[0112] The energy scheduling screening module is used to formulate an energy scheduling plan based on the prediction results of future environmental data through a multi-objective optimization algorithm, determine the energy usage strategy of the heating, cooling and lighting systems, and output the optimal energy scheduling plan.

[0113] Among them, environmental data include temperature, humidity, light intensity, carbon dioxide concentration and nutrient solution flow rate, energy supply data include solar energy and electrical energy, preprocessing includes missing value processing, outlier detection and data smoothing, the gas exchange model is used to simulate the dynamic changes of carbon dioxide concentration and oxygen in the greenhouse, and the kinetic nutrient transfer model is used to simulate the transmission and distribution of nutrient solution in the hydroponic system.

[0114] In this embodiment, the data preprocessing module of the power demand response system scheduling device based on model prediction provided by the present invention is specifically used for: Environmental data inside and outside the greenhouse, crop growth demand data, and energy supply data are monitored in real time through multiple sensors and energy monitoring equipment, and original data sets are constructed based on the environmental data, crop growth demand data, and energy supply data.

[0115] The missing values ​​of the original data set are filled with the mean of the corresponding data column in the original data set as the filling value, and the three sigma principle is used to detect outliers in the original data set, so that when an outlier is detected, the outlier is replaced with the median of the corresponding column in the original data set.

[0116] The original data set is smoothed by the moving average method, and the cleaned environmental data, crop growth demand data, and energy supply data are integrated to output an integrated data set.

[0117] In this embodiment, in the power demand response system scheduling device based on model prediction provided by the present invention, the data preprocessing module is specifically used for: The integrated data set is standardized and normalized to eliminate the influence of different dimensions and magnitudes in the integrated data set on the gas exchange model and the kinetic nutrient transfer model, and the converted data set is output.

[0118] The standardized processing expression is: The normalized expression is: In the formula, and are the i-th row and j-th list elements respectively The standardized and normalized results of is the element in the i-th row and j-th column of the integrated data set, is the mean of the jth column, is the standard deviation of the jth column, and are the minimum and maximum values ​​of the jth column respectively.

[0119] In this embodiment, the model building module of the power demand response system scheduling device based on model prediction provided by the present invention is specifically used for: The first basic parameters required for the gas exchange model are obtained from the converted data set, the first basic parameters including the carbon dioxide concentration, oxygen concentration, ventilation rate and photosynthesis rate inside and outside the greenhouse, and the first basic data set of the gas exchange model is constructed.

[0120] Based on the first basic data set, a kinetic gas exchange equation is established to describe the change of carbon dioxide concentration and oxygen concentration over time in the greenhouse. The expression of the kinetic gas exchange equation is: In the formula, is the rate of change of carbon dioxide concentration over time, is the rate of change of oxygen concentration over time, is the total volume of greenhouse gases, is the ventilation rate, is the photosynthesis rate, is the concentration of carbon dioxide entering the greenhouse, is the oxygen concentration entering the greenhouse, and are the carbon dioxide concentration and oxygen concentration in the greenhouse at time t, is the total photosynthetic area of ​​the crop at time t.

[0121] In this embodiment, the model building module of the power demand response system scheduling device based on model prediction provided by the present invention is specifically used for: The second basic parameters required for the dynamic nutrient transfer model are obtained from the converted data set. The second basic parameters include nutrient solution flow, nutrient concentration in the nutrient solution, total plant root area, and nutrient absorption rate. The second basic parameter set is constructed. Based on the second basic parameter set, the kinetic nutrient transfer equation describing the nutrient changes in the nutrient solution is established. The expression of the kinetic nutrient transfer equation is: In the formula, is the rate of change of nutrient concentration over time, is the volume of the nutrient solution storage tank, is the nutrient concentration of the input nutrient solution, is the total root area of ​​the plant, is the nutrient solution flow rate, is the nutrient concentration in the nutrient solution at time t, is the nutrient absorption rate.

[0122] The parameters of the gas exchange model and the kinetic nutrient transfer model are calibrated by minimizing the error between the model prediction value and the actual monitoring value, and the first basic parameter set and the second basic parameter set are divided into a training set and a validation set by cross-validation or holdout method. The determination coefficient and mean square error index are calculated to verify the gas exchange model and the kinetic nutrient transfer model, and the calibrated model output parameters and status are output.

[0123] In this embodiment, the future environment prediction module of the power demand response system scheduling device based on model prediction provided by the present invention is specifically used for: A Bayesian network training data set is constructed based on the calibrated model output parameters and states, and the dependency relationship between various environmental data in the Bayesian network is determined according to the Bayesian network training data set through a structural learning algorithm to construct a Bayesian network structure.

[0124] The maximum likelihood estimation or Bayesian estimation algorithm is used to calculate the conditional probability table of each node according to the Bayesian network training data set, and the Bayesian network is trained in combination with the Bayesian network structure and the Bayesian network training data set. The model structure and parameters are adjusted through cross-validation to obtain the trained Bayesian network model.

[0125] Among them, the Bayesian network structure consists of a node set and an edge set. The node set represents each environmental data, and the edge set represents the directed dependency relationship between nodes. The structural learning algorithm includes a scoring search algorithm and a constraint-based algorithm. The trained Bayesian network model is used to predict and output the prediction results of future environmental data based on current or historical environmental data.

[0126] In this embodiment, the present invention provides a model-prediction-based electricity demand response system scheduling device, and the multi-objective optimization algorithm includes an optimization objective function and a predefined set of constraints. The optimization objective function contains multiple sub-objectives, and the sub-objectives include minimizing energy costs, maximizing energy utilization efficiency, and ensuring environmental stability.

[0127] The energy scheduling screening module is specifically used for: Based on the weight coefficient of each sub-goal in the optimization objective function, the sub-goals are weighted and summed to obtain the comprehensive optimization objective function. Combined with the constraint set, the configuration parameters are determined by the non-dominated sorting genetic algorithm to construct a configured multi-objective optimization algorithm.

[0128] The configured multi-objective optimization algorithm is called to generate a solution set that approximates the Pareto optimal solution based on the comprehensive optimization objective function and constraint set, and the optimal energy scheduling solution is selected from the solution set according to actual needs and priorities.

[0129] The configuration parameters include population size, number of iterations, crossover rate, and mutation rate. Each solution in the solution set corresponds to an energy scheduling plan.

[0130] The present disclosure may be a system, a method and / or a computer program product. The computer program product may include a computer-readable storage medium carrying computer-readable program instructions for causing a processor to implement various aspects of the present disclosure.

[0131] A computer-readable storage medium may be a tangible device that can hold and store instructions used by an instruction execution device. A computer-readable storage medium may be, for example, but not limited to, an electrical storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination of the above. More specific examples (a non-exhaustive list) of computer-readable storage media include: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a static random access memory (SRAM), a portable compact disk read-only memory (CD-ROM), a digital versatile disk (DVD), a memory stick, a floppy disk, a mechanical encoding device, such as a punch card or a raised structure in a groove on which instructions are stored, and any suitable combination of the above. The computer-readable storage medium used herein is not to be interpreted as a transient signal itself, such as a radio wave or other freely propagating electromagnetic wave, an electromagnetic wave propagating through a waveguide or other transmission medium (e.g., a light pulse through a fiber optic cable), or an electrical signal transmitted through a wire.

[0132] The computer-readable program instructions described herein can be downloaded from a computer-readable storage medium to each computing / processing device, or downloaded to an external computer or external storage device via a network, such as the Internet, a local area network, a wide area network, and / or a wireless network. The network can include copper transmission cables, optical fiber transmissions, wireless transmissions, routers, firewalls, switches, gateway computers, and / or edge servers. The network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards the computer-readable program instructions for storage in the computer-readable storage medium in each computing / processing device.

[0133] The computer program instructions for performing the operations of the present disclosure may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, state setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages, such as Smalltalk, C++, etc., and conventional procedural programming languages, such as "C" language or similar programming languages. The computer-readable program instructions may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer, partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., using an Internet service provider to connect through the Internet). In some embodiments, by using the state information of the computer-readable program instructions to personalize an electronic circuit, such as a programmable logic circuit, a field programmable gate array (FPGA), or a programmable logic array (PLA), the electronic circuit may execute the computer-readable program instructions, thereby implementing various aspects of the present disclosure.

[0134] Various aspects of the present disclosure are described herein with reference to flowcharts and / or block diagrams of methods, devices (systems) and computer program products according to embodiments of the present disclosure. It should be understood that each box in the flowchart and / or block diagram and the combination of boxes in the flowchart and / or block diagram can be implemented by computer-readable program instructions.

[0135] These computer-readable program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, thereby producing a machine, so that when these instructions are executed by the processor of the computer or other programmable data processing device, a device that implements the functions / actions specified in one or more boxes in the flowchart and / or block diagram is generated. These computer-readable program instructions can also be stored in a computer-readable storage medium, and these instructions cause the computer, programmable data processing device, and / or other equipment to work in a specific manner, so that the computer-readable medium storing the instructions includes a manufactured product, which includes instructions for implementing various aspects of the functions / actions specified in one or more boxes in the flowchart and / or block diagram.

[0136] Computer-readable program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other device so that a series of operating steps are performed on the computer, other programmable data processing apparatus, or other device to produce a computer-implemented process, thereby causing the instructions executed on the computer, other programmable data processing apparatus, or other device to implement the functions / actions specified in one or more boxes in the flowchart and / or block diagram.

[0137] The flow chart and block diagram in the accompanying drawings show the possible architecture, function and operation of the system, method and computer program product according to multiple embodiments of the present disclosure. In this regard, each square box in the flow chart or block diagram can represent a part of a module, program segment or instruction, and a part of the module, program segment or instruction includes one or more executable instructions for realizing the specified logical function. In some alternative implementations, the function marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two continuous square boxes can actually be executed substantially in parallel, and they can sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs the specified function or action, or can be implemented with a combination of special hardware and computer instructions.

[0138] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents, and any modifications or equivalent replacements that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for dispatching power demand response based on a dynamic prediction model, characterized in that: The method comprises: Acquire environmental data inside and outside the greenhouse, crop growth demand data, and energy supply data in real time, and pre-process the environmental data, crop growth demand data, and energy supply data to construct a data set; Constructing a gas exchange model and a kinetic nutrient transfer model based on the data set, and outputting parameters and states of the gas exchange model and the kinetic nutrient transfer model; Analyze the parameters and states output by the gas exchange model and the kinetic nutrient transfer model through a Bayesian network to predict the change trend of the environmental data in the greenhouse in the first time period in the future, and output the prediction results of the future environmental data; Formulate an energy scheduling plan based on the prediction results of the future environmental data through a multi-objective optimization algorithm, determine the energy use strategy of the heating, cooling and lighting systems, and output the optimal energy scheduling plan; Among them, the environmental data includes temperature, humidity, light intensity, carbon dioxide concentration and nutrient solution flow rate, the energy supply data includes solar energy and electrical energy, the preprocessing includes missing value processing, outlier detection and data smoothing processing, the gas exchange model is used to simulate the dynamic changes of carbon dioxide concentration and oxygen in the greenhouse, and the kinetic nutrient transfer model is used to simulate the transmission and distribution of nutrient solution in the hydroponic system.

2. The power demand response scheduling method based on the dynamic prediction model according to claim 1 is characterized in that: The real-time acquisition of environmental data inside and outside the greenhouse, crop growth demand data, and energy supply data, and preprocessing of the environmental data, crop growth demand data, and energy supply data to construct a data set includes: Monitoring the environmental data, crop growth demand data and energy supply data inside and outside the greenhouse in real time through multiple sensors and energy monitoring equipment, and constructing an original data set based on the environmental data, crop growth demand data and energy supply data; Filling missing values ​​in the original data set with the mean of the corresponding data column in the original data set as the filling value, and detecting outliers in the original data set using the three sigma principle, so that when an outlier is detected, the outlier is replaced with the median of the corresponding column in the original data set; The raw data set is smoothed by a moving average method, and the cleaned environmental data, crop growth demand data, and energy supply data are integrated to output an integrated data set.

3. The power demand response scheduling method based on the dynamic prediction model according to claim 2 is characterized in that: The real-time acquisition of environmental data inside and outside the greenhouse, crop growth demand data, and energy supply data, and preprocessing the environmental data, crop growth demand data, and energy supply data to construct a data set also includes: Standardizing and normalizing the integrated data set to eliminate the influence of different dimensions and magnitudes in the integrated data set on the gas exchange model and the kinetic nutrient transfer model, and outputting the converted data set; The standardized processing expression is: ； The normalized expression is: ； In the formula, and are the i-th row and j-th list elements respectively The standardized and normalized results of is the element in the i-th row and j-th column of the integrated data set, is the mean of the jth column, is the standard deviation of the jth column, and are the minimum and maximum values ​​of the jth column respectively.

4. The power demand response scheduling method based on the dynamic prediction model according to claim 3 is characterized in that: The step of constructing a gas exchange model and a kinetic nutrient transfer model based on the data set, and outputting parameters and states of the gas exchange model and the kinetic nutrient transfer model, includes: Acquire the first basic parameters required by the gas exchange model from the converted data set, the first basic parameters including the carbon dioxide concentration, oxygen concentration, ventilation rate and photosynthesis rate inside and outside the greenhouse, and construct the first basic data set of the gas exchange model; Based on the first basic data set, a kinetic gas exchange equation is established to describe the change of carbon dioxide concentration and oxygen concentration in the greenhouse over time. The expression of the kinetic gas exchange equation is: ， ； In the formula, is the rate of change of carbon dioxide concentration over time, is the rate of change of oxygen concentration over time, is the total volume of greenhouse gases, is the ventilation rate, is the photosynthesis rate, is the concentration of carbon dioxide entering the greenhouse, is the oxygen concentration entering the greenhouse, and are the carbon dioxide concentration and oxygen concentration in the greenhouse at time t, is the total photosynthetic area of ​​the crop at time t.

5. The power demand response scheduling method based on the dynamic prediction model according to claim 4 is characterized in that: The step of constructing a gas exchange model and a kinetic nutrient transfer model based on the data set, and outputting parameters and states of the gas exchange model and the kinetic nutrient transfer model, further includes: Acquire the second basic parameters required by the kinetic nutrient transfer model from the converted data set, the second basic parameters including nutrient solution flow rate, nutrient concentration in the nutrient solution, total plant root area and nutrient absorption rate, and construct a second basic parameter set; Based on the second basic parameter set, a kinetic nutrient transfer equation describing the nutrient changes in the nutrient solution is established. The expression of the kinetic nutrient transfer equation is: ； In the formula, is the rate of change of nutrient concentration over time, is the volume of the nutrient solution storage tank, is the nutrient concentration of the input nutrient solution, is the total root area of ​​the plant, is the nutrient solution flow rate, is the nutrient concentration in the nutrient solution at time t, is the nutrient absorption rate; The gas exchange model and the kinetic nutrient transfer model are calibrated by minimizing the error between the model prediction value and the actual monitoring value, and the first basic parameter set and the second basic parameter set are divided into a training set and a validation set by cross-validation or holdout method, and the determination coefficient and the mean square error index are calculated to verify the gas exchange model and the kinetic nutrient transfer model, and the calibrated model output parameters and status are output.

6. The power demand response scheduling method based on the dynamic prediction model according to claim 5 is characterized in that: The Bayesian network is used to analyze the parameters and states output by the gas exchange model and the kinetic nutrient transfer model to predict the change trend of the environmental data in the greenhouse in the first time period in the future, and output the prediction results of the future environmental data, including: Constructing a Bayesian network training data set based on the calibrated model output parameters and states, and determining the dependency between the environmental data in the Bayesian network according to the Bayesian network training data set through a structural learning algorithm to construct a Bayesian network structure; The conditional probability table of each node is calculated according to the Bayesian network training data set using a maximum likelihood estimation or Bayesian estimation algorithm, and the Bayesian network is trained in combination with the Bayesian network structure and the Bayesian network training data set, and the model structure and parameters are adjusted through cross-validation to obtain a trained Bayesian network model; Among them, the Bayesian network structure consists of a node set and an edge set, the node set represents each environmental data, the edge set represents the directed dependency relationship between nodes, the structural learning algorithm includes a scoring search algorithm and a constraint-based algorithm, and the trained Bayesian network model is used to predict and output the prediction results of the future environmental data based on current or historical environmental data.

7. The power demand response scheduling method based on the dynamic prediction model according to claim 6 is characterized in that: The multi-objective optimization algorithm includes an optimization objective function and a predefined set of constraints, wherein the optimization objective function includes multiple sub-objectives, wherein the sub-objectives include minimizing energy costs, maximizing energy utilization efficiency, and ensuring environmental stability; The energy scheduling scheme is formulated according to the prediction results of the future environmental data by the multi-objective optimization algorithm, and the energy use strategy of the heating, cooling and lighting systems is determined to output the optimal energy scheduling scheme, including: Based on the weight coefficient of each sub-objective in the optimization objective function, the sub-objectives are weighted and summed to obtain a comprehensive optimization objective function, and the configuration parameters are determined by a non-dominated sorting genetic algorithm in combination with the constraint condition set to construct a configured multi-objective optimization algorithm; Calling the configured multi-objective optimization algorithm to generate a solution set of approximate Pareto optimal solutions based on the comprehensive optimization objective function and the constraint condition set, and screening out the optimal energy scheduling solution from the solution set according to actual needs and priorities; The configuration parameters include population size, number of iterations, crossover rate and mutation rate, and each solution in the solution set corresponds to an energy scheduling scheme.

8. A power demand response system dispatching device based on model prediction, characterized in that: The device comprises: A data preprocessing module, used to obtain environmental data inside and outside the greenhouse, crop growth demand data, and energy supply data in real time, and preprocess the environmental data, crop growth demand data, and energy supply data to construct a data set; A model building module, used to build a gas exchange model and a kinetic nutrient transfer model based on the data set, and output parameters and states of the gas exchange model and the kinetic nutrient transfer model; A future environment prediction module is used to analyze the parameters and states output by the gas exchange model and the kinetic nutrient transfer model through a Bayesian network to predict the change trend of the environmental data in the greenhouse within the first time period in the future and output the prediction results of the future environmental data; An energy scheduling screening module is used to formulate an energy scheduling plan according to the prediction results of the future environmental data through a multi-objective optimization algorithm, determine the energy use strategy of the heating, cooling and lighting systems, and output the optimal energy scheduling plan; Among them, the environmental data includes temperature, humidity, light intensity, carbon dioxide concentration and nutrient solution flow rate, the energy supply data includes solar energy and electrical energy, the preprocessing includes missing value processing, outlier detection and data smoothing processing, the gas exchange model is used to simulate the dynamic changes of carbon dioxide concentration and oxygen in the greenhouse, and the kinetic nutrient transfer model is used to simulate the transmission and distribution of nutrient solution in the hydroponic system.

9. A terminal comprising a processor and a storage medium; characterized in that: The storage medium is used to store instructions; The processor is configured to operate according to the instructions to execute the steps of the method according to any one of claims 1-7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images