Greenhouse micro-grid dynamic scheduling and photosynthetic efficiency collaborative optimization control method and system

Abstract

The invention relates to the technical field of facility agriculture and micro-grid cooperative control, and particularly discloses a greenhouse micro-grid dynamic scheduling and photosynthetic efficiency cooperative optimization control method and system. The method comprises the following steps: constructing an energy management scheduling model with maximization of accumulated net photosynthetic products of crops in a scheduling period and minimization of net energy cost of a micro-grid as a collaborative optimization target; collecting real-time environment parameter data in the greenhouse and operation data of a micro-grid wind and light storage system and a load; a multi-objective optimization algorithm is adopted to solve the model online, and a joint optimization strategy including a micro-grid dynamic scheduling instruction and a greenhouse environment coordination control instruction is output; and executing the joint optimization strategy, and dynamically regulating and controlling the power output distribution of the micro-grid, the energy storage charging and discharging state and the operation process of the environment regulation and control equipment. According to the method, the multi-target collaborative optimization model with'photosynthetic product accumulation maximization 'and'net energy cost minimization' as the core is constructed, so that comprehensive maximization of energy benefits and agricultural production benefits is realized.

Description

A method and system for dynamic scheduling and coordinated optimization control of photosynthetic efficiency in greenhouse microgrids Technical Field

[0001] This invention relates to the field of coordinated control technology of facility agriculture and microgrids, specifically to a method and system for dynamic scheduling and coordinated optimization control of photosynthetic efficiency of greenhouse microgrids. Background Technology

[0002] With the rapid development of facility agriculture, greenhouses, as efficient agricultural production platforms, require significant energy consumption for internal environmental control (such as light, temperature, humidity, and carbon dioxide concentration regulation). To reduce energy costs and improve energy efficiency, wind-solar-storage microgrid systems are widely used in greenhouse applications. However, in existing technologies, greenhouse microgrid scheduling and agricultural production control are often carried out independently, leading to the following problems:

[0003] The core objective of microgrid dispatch is often to minimize energy costs or maximize energy efficiency, without fully considering the photosynthetic efficiency requirements of crops, leading to a mismatch between energy allocation and crop growth needs. For example, during peak crop photosynthetic demand periods, microgrids may prioritize energy storage or electricity sales rather than supplying power to supplemental lighting equipment, thus affecting the accumulation of photosynthetic products in crops. Conversely, during peak wind and solar power generation periods, excess power generation may result in energy waste, and the surplus energy may not be used to optimize the crop growth environment.

[0004] Crop growth environment regulation is typically implemented based on fixed growth thresholds, failing to dynamically adjust according to the real-time output status of the microgrid. Therefore, it cannot achieve a coordinated match between energy consumption and microgrid output. For example, when microgrid output is insufficient, high-energy-consuming environmental control equipment continues to operate in conventional mode, leading to a surge in electricity purchase costs; conversely, when microgrid output is sufficient, environmental parameters are not adjusted in time to improve photosynthetic efficiency, wasting energy utilization potential.

[0005] Most existing collaborative control technologies focus only on matching a single environmental factor (such as sunlight) with photovoltaic output. They do not comprehensively consider the synergistic effects of multiple environmental factors such as temperature, humidity, and carbon dioxide concentration on crop photosynthetic efficiency, nor do they construct multi-objective collaborative optimization models, making it difficult to achieve the comprehensive maximization of power generation revenue and crop yield revenue.

[0006] In the prior art, Chinese patent CN120749875A discloses an adaptive power generation optimization control system for greenhouse photovoltaic roofs, the core of which lies in adjusting the tilt angle of photovoltaic panels to balance crop light demand and power generation efficiency. However, this technology is limited to adjusting the tilt angle of the photovoltaic roof and does not involve the overall dynamic scheduling of wind-solar-storage microgrids; at the same time, it does not comprehensively consider the impact of multiple environmental factors such as temperature, humidity, and carbon dioxide concentration on photosynthetic efficiency, resulting in a relatively simple collaborative control dimension. Chinese patent CN119067387A discloses a virtual power plant scheduling method for facility agriculture, the core of which is to achieve optimal allocation of energy resources in facility agriculture parks through hierarchical cluster control, aiming to reduce operating costs and improve environmental benefits. However, this technology focuses on the interactive scheduling between the virtual power plant and the main grid, and does not design for the dynamic characteristics of the microgrid inside a single greenhouse and the precise coordination with crop photosynthetic efficiency, thus failing to meet the needs of refined collaborative control in the scenario of a single greenhouse.

[0007] Therefore, there is an urgent need to propose a control method and system that can achieve dynamic scheduling of greenhouse microgrids and coordinated optimization of crop photosynthetic efficiency, so as to balance the power output of microgrids and the demand for crop photosynthetic efficiency, and ultimately maximize the comprehensive benefits of power generation and crop yield. Summary of the Invention

[0008] The purpose of this invention is to overcome the shortcomings of existing technologies, namely the independent and poor coordination between greenhouse microgrid scheduling and crop photosynthetic efficiency regulation, and to provide a method and system for dynamic scheduling and coordinated optimization control of greenhouse microgrid and photosynthetic efficiency, thereby achieving synergistic efficiency improvement in energy regulation and agricultural production.

[0009] To achieve the above objectives, the present invention adopts the following technical solution:

[0010] A method for dynamic scheduling and synergistic optimization control of photosynthetic efficiency in greenhouse microgrids includes the following steps:

[0011] S1: Construct a multi-objective energy management and scheduling model with the goal of maximizing the cumulative net photosynthetic products of crops and minimizing the net energy cost of the microgrid within the scheduling period M, while simultaneously satisfying the microgrid operation constraints and crop growth environment constraints.

[0012] S2: Real-time acquisition of system status data This includes spatial distribution data of greenhouse internal environmental parameters, crop growth status data, and distributed power generation output, energy storage status of charge, load demand, and grid-connected power purchase and sale data in the microgrid; and prediction sequences of environmental parameters, wind and solar power generation output, and load demand within the future scheduling period [k+1, k+M-1] generated by the prediction model. ;

[0013] S3: Real-time data collected in step S2 and predicted sequence A multi-objective optimization algorithm is used to solve the model constructed in step S1 online, and the result is a joint optimization strategy sequence that includes both microgrid dynamic scheduling instructions and greenhouse environment coordinated control instructions for the next scheduling cycle. ;

[0014] S4: Execute the control command corresponding to the current time in the joint optimization strategy sequence. After a preset scheduling interval, at time k+1, the process returns to step S2, regenerates the prediction sequence based on the latest collected real-time data, repeats the optimization solution in step S3, and executes the process in step S4, thereby achieving closed-loop feedback and rolling optimization.

[0015] Preferably, in step S1, the objective function of the multi-objective energy management and scheduling model is constructed as follows: objective function F = ω1 × F1 + ω2 × F2, where F1 is the objective function for maximizing cumulative photosynthetic products, F2 is the objective function for minimizing net energy costs, and ω1 and ω2 are weight coefficients that satisfy ω1 + ω2 = 1; F1 is constructed based on the crop photosynthetic rate model, and its expression is:

[0016]

[0017] In the formula, M is the scheduling cycle, S is the greenhouse planting area, and P is the greenhouse planting area. n The net photosynthetic rate of the crop is given by the Farquhar model, where I(t,s) is the light intensity at time t and location s inside the greenhouse, T(t,s) is the temperature at time t and location s inside the greenhouse, C(t,s) is the carbon dioxide concentration at time t and location s inside the greenhouse, H(t,s) is the humidity at time t and location s inside the greenhouse, and k is the net photosynthetic rate of the crop. yp To convert the dimensions of F1 and F2 so that their dimensions are consistent, the expression for F2 is:

[0018]

[0019] in:

[0020] , These represent the electricity purchased and sold by the microgrid to the external power grid, respectively.

[0021] , , , These represent the power generation capacity of photovoltaic and wind power, and the charging and discharging power of energy storage batteries, respectively.

[0022] , , , These are the unit operating costs of photovoltaic, wind power, energy storage batteries, and environmental equipment, respectively.

[0023] , These represent the unit costs of purchasing and selling electricity from the external power grid to the microgrid, respectively.

[0024] Power consumption of environmental equipment, such as supplemental lighting, heating devices, and CO2 generators;

[0025] The power generation capacity of wind and solar power generation equipment and the charging and discharging capacity of energy storage batteries.

[0026] The decision variables of the objective function are: ,in, For continuous power variables, Set point variables for the environment. A binary variable representing the on / off mode of device operation.

[0027] The constraints of the multi-objective energy management and scheduling model include microgrid operation constraints and crop growth environment constraints.

[0028] The microgrid operation constraints include at least power balance constraints, wind and solar power output constraints, energy storage device charging and discharging constraints, and electricity purchase and sale constraints.

[0029] The power balance constraint is as follows:

[0030] ;

[0031] The power output constraints for wind and solar power generation are: ,

[0032] The charging and discharging constraints of energy storage devices are:

[0033] ;

[0034] The restrictions on the purchase and sale of electricity are as follows: , ;

[0035] The constraints on the crop growth environment are as follows: In the formula, I min,i I max,i These represent the minimum and maximum suitable light intensity during the growth period of crop i, respectively, T. min,i T max,i These represent the minimum and maximum suitable temperatures during the growth period of crop i, H.min,i H max,i These represent the minimum and maximum suitable humidity values ​​during the growth period of crop i, respectively. min,i C max,i These represent the minimum and maximum suitable carbon dioxide concentrations during the growth period of crop i, respectively.

[0036] Preferably, the multi-objective optimization algorithm in step S3 is a non-dominated sorting genetic algorithm (NSGA-III), and the improvement of this algorithm is as follows: the crossover probability and mutation probability are dynamically adjusted based on the priority of crop photosynthetic efficiency, and a microgrid operation stability constraint factor is introduced to optimize the fitness function; its online solution process specifically includes:

[0037] S31: Population Initialization: Randomly generate N initial solutions that satisfy all constraints within the feasible region. Each solution set includes microgrid scheduling variables and environmental control variables; the microgrid scheduling variables include wind and solar power generation output allocation coefficients, energy storage device charging and discharging power, and power purchased and sold with the external grid; the environmental control variables include the setting parameters of supplemental lighting, temperature control devices, carbon dioxide generators, and humidifiers.

[0038] S32: Dynamic parameter adjustment:

[0039] For the g-th generation population P g Based on the current crop growth period photosynthetic efficiency priority factor β g The crossover probability P of the dynamic adjustment algorithm c With the probability of mutation P m The strategy is adjusted as follows:

[0040] , ;

[0041] in, , , and To adjust the constant and ensure that P is appropriately increased during the peak photosynthetic demand period of crops. c And reduce P m ;

[0042] S33: Fitness Assessment: Calculate the fitness of each individual in the current population. fitness value Where α is the microgrid operation stability constraint factor, and its value is related to the power fluctuation between adjacent scheduling periods. Negative correlation, the strategy is: ,in This indicates that the average value over the scheduling period is used to penalize scheduling schemes that cause drastic power fluctuations.

[0043] S34: Iterative Evolution: The population is clustered based on non-dominated sorting and reference vectors, using tournament selection; simulated binary crossover (SBX) is used for continuous variables, and uniform crossover is used for discrete variables, with probabilities of... For continuous variables, polynomial mutation (PM) is used; for discrete variables, bit-flip mutation is used, with a probability of... ;

[0044] S35: Convergence Judgment and Output: Repeat steps S32 to S34, until g ≥ G is satisfied. max (Maximum number of iterations) or (HV is the hypervolume index) The process terminates when the population reaches a certain threshold. From the non-dominated solution set of the final generation, a compromise solution chr is selected based on linear scalarization using weight coefficients ω1 and ω2 according to a preset preference. ∗ The decoded output is a joint optimization strategy U that includes specific power values ​​and environmental device settings. ∗ .

[0045] Preferably, in step S2, spatial distribution data of greenhouse environmental parameters are collected through a distributed wireless sensor network; crop types, growth stages, and leaf area index are identified through a machine vision system; real-time power data are collected through a microgrid energy management system; the prediction model includes a wind and solar power output prediction model, a greenhouse environment prediction model, and a load demand prediction model.

[0046] Preferably, the microgrid dynamic scheduling command is based on decision variables of the objective function, including the output setpoints of the photovoltaic inverter and wind turbine converter, the charging and discharging power command of the energy storage converter, and the power exchange command of the grid connection point connected to the external power grid; the environmental coordinated control command includes the setpoints of the supplementary lighting intensity, the set temperature of the air conditioning system, the CO2 generator rate, and the humidifier power.

[0047] A dynamic scheduling and photosynthetic efficiency synergistic optimization control system for greenhouse microgrids, the system having a hierarchical structure, comprising:

[0048] Perception and Prediction Layer: Composed of distributed IoT nodes, machine vision units, and state estimators, responsible for implementing S2 functions;

[0049] Optimized decision layer: An improved evolutionary algorithm is run by a high-performance embedded computing unit, which is responsible for implementing the S3 function;

[0050] The execution control layer consists of a programmable logic controller and a power converter drive circuit, and is responsible for implementing the S4 function;

[0051] Each layer interacts with other layers via a real-time data bus, forming a closed loop.

[0052] Preferably, the data acquisition and prediction module includes an image acquisition unit, a distributed sensor network unit, a microgrid data monitoring unit, and a data processing server connected to the above units for running the prediction model;

[0053] The image acquisition unit includes a high-definition camera and an image processing module. The high-definition camera is used to acquire crop images, and the image processing module is used to analyze the crop images to obtain crop growth status data.

[0054] The distributed sensor network unit includes sensor nodes and gateways evenly deployed in different areas of the greenhouse. The sensor nodes integrate various sensors for collecting environmental parameter data in different areas of the greenhouse.

[0055] The microgrid data monitoring unit is used to collect data in real time on the output of wind and solar power generation equipment, the charging and discharging status of energy storage equipment, load power, and power purchased and sold.

[0056] Preferably, the execution control module includes a microgrid control unit and an environmental control unit;

[0057] The microgrid control unit is connected to the wind and solar power generation equipment, energy storage equipment, and grid-connected switch in the microgrid, and is used to execute dynamic dispatch commands of the microgrid.

[0058] The environmental control unit is connected to the supplemental lighting, temperature control device, ventilation device, CO2 generator, and humidifier inside the greenhouse, and is used to execute the greenhouse environment coordinated control commands.

[0059] Compared with the prior art, the present invention has the following beneficial effects:

[0060] A multi-objective synergistic optimization model centered on "maximizing the accumulation of photosynthetic products" and "minimizing net energy costs" was constructed, overcoming the limitations of single-objective regulation in existing technologies and achieving comprehensive maximization of energy and agricultural production benefits. In existing technologies, CN120749875A only balances light intensity and power generation efficiency by adjusting the angle of photovoltaic panels, without addressing energy cost optimization; CN119067387A focuses on the economic dispatch and environmental benefits of a virtual power plant, without considering the core requirement of crop photosynthetic product accumulation. In contrast, this invention achieves synergistic optimization of both objectives through multi-objective integration.

[0061] This invention comprehensively considers the synergistic effects of multiple environmental factors, such as light intensity, temperature, humidity, and CO2 concentration, on crop photosynthetic efficiency, and constructs an objective function for the accumulation of photosynthetic products based on net photosynthetic rate. Compared to existing methods that only focus on light intensity, the proposed method offers higher control precision and better meets the actual needs of crop growth.

[0062] An improved collaborative optimization intelligent algorithm was proposed. By dynamically adjusting the algorithm parameters and optimizing the fitness function, the algorithm balances the priority of crop photosynthetic efficiency with the stability of microgrid operation, thus ensuring the scientific nature and feasibility of the scheduling instructions.

[0063] A comprehensive collaborative control system was constructed, realizing closed-loop control of data acquisition, model solving, and command execution. This system can dynamically adjust control strategies based on real-time microgrid output and crop growth status. Compared to existing static control methods, this invention is more adaptable and can achieve synergistic effects between energy regulation and agricultural production. Attached Figure Description

[0064] Figure 1 is a flowchart of the method of the present invention;

[0065] Figure 2 is a schematic diagram of the module structure of the system of the present invention. Detailed Implementation

[0066] The present application will now be described in further detail with reference to the accompanying drawings and embodiments. It is to be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of the application.

[0067] Example 1: Specific application of the method of dynamic scheduling and synergistic optimization control of photosynthetic efficiency in greenhouse microgrids

[0068] This embodiment uses a modern glass greenhouse and its supporting distributed wind-solar-storage microgrid as an application scenario to illustrate the implementation process of the invention in detail. The greenhouse has dimensions of 50m × 20m (planting area S = 1000 m²) and adopts a north-south orientation to improve light energy utilization efficiency. The greenhouse is equipped with a wind-solar-storage microgrid system, which includes a photovoltaic array, a small wind turbine, and a lithium-ion battery energy storage system, and can achieve bidirectional energy interaction with the public power grid. The selected crop is the main greenhouse variety "Zhongza 105" tomato, with a planting density of 3000 plants / acre. The experimental period covers the entire tomato growth cycle (from planting to harvest, a total of 150 days). This embodiment achieves synergistic optimization of microgrid scheduling and photosynthetic efficiency through a complete process of "data acquisition—model construction—optimization solution—execution control—effect verification". The specific steps are as follows:

[0069] S1: Construct a multi-objective energy management and scheduling model for a greenhouse wind-solar-storage microgrid and calibrate its parameters.

[0070] 1.1 Determine the photosynthetic production target F1

[0071] At each decision time k, construct a mixed integer nonlinear programming (MINLP) problem covering M time periods of the future scheduling cycle (in this embodiment, M=24, the scheduling interval Δt=1 hour, that is, the scheduling cycle is 24 hours).

[0072] The overall objective function is F = ω1 × F1 + ω2 × F2, where F1 is the objective function for maximizing cumulative photosynthetic products, F2 is the objective function for minimizing net energy costs, and ω1 and ω2 are weight coefficients that satisfy ω1 + ω2 = 1.

[0073] The day is divided into peak photosynthetic periods (6:00–18:00) and non-peak photosynthetic periods (18:00–6:00 the next day), with different optimization weights assigned to each. The transplanting period (days 1–30) and fruit enlargement period (days 90–120) are critical periods for photosynthetic demand, with weighting coefficients set to ω1=0.6 and ω2=0.4. During the vegetative growth period (days 31–60) and harvest period (days 121–150), the weighting coefficients are adjusted to ω1=0.5 and ω2=0.5. The objective function is to maximize the accumulation of photosynthetic products.

[0074]

[0075] The integral is converted into a summation form that is easier for computers to solve:

[0076]

[0077] in, k is the number of spatial grids. yp It was estimated based on the total yield, total F1, and tomato unit price throughout the entire planting period.

[0078] 1.2 Calibration of Photosynthetic Rate Model Parameters

[0079] Net photosynthetic rate P of tomatoes n A modified Farquhar model was used, with parameters calibrated based on previous field trial data. Using a LI-6800 portable photosynthesis measurement system, the net photosynthetic rate of tomatoes was measured under different combinations of light intensity, temperature, humidity, and CO2 concentration. A total of 300 sets of valid data were collected, and the model parameters, including the maximum photosynthetic rate P, were obtained through least-squares fitting. max =28μmol / m² / s, optical response coefficient k=0.005, CO2 compensation point Γ=50ppm, Michaelis constant K n =200ppm. The humidity effect factor f(H) is defined as follows: when H∈[60%, 80%], f(H)=1; when H<60%, f(H)=0.025H−0.5; when H>80%, f(H)=−0.025H+3 (ensuring f(H)=0.75 when H=50% and f(H)=0.75 when H=90% to reflect the inhibitory effect of excessively low or high humidity on photosynthetic rate). The final determined expression for the net photosynthetic rate of tomato is as follows:

[0080]

[0081] 1.3 Determine the energy target F2

[0082] The specific formula for calculating F2 is described below:

[0083]

[0084] in:

[0085] , These represent the electricity purchased and sold by the microgrid to the external power grid, respectively.

[0086] , , , These represent the power generation capacity of photovoltaic and wind power, and the charging and discharging power of energy storage batteries, respectively.

[0087] , , , These are the unit operating costs of photovoltaic, wind power, energy storage batteries, and environmental equipment, respectively.

[0088] , These represent the unit costs of purchasing and selling electricity from the external power grid to the microgrid, respectively.

[0089] Power consumption of environmental equipment, such as supplemental lighting, heating devices, and CO2 generators;

[0090] The power generation capacity of wind and solar power generation equipment and the charging and discharging capacity of energy storage batteries.

[0091] Electricity purchase price is set according to the time-of-use pricing policy. The electricity price is as follows: Peak hours (8:00–12:00, 17:00–21:00) 0.85 yuan / kWh; Average hours (6:00–8:00, 12:00–17:00, 21:00–24:00) 0.52 yuan / kWh; Off-peak hours (0:00–6:00) 0.28 yuan / kWh. Electricity sales price. The price is uniformly set at 0.35 yuan / kWh, which is in line with the local grid-connected electricity price standard for distributed photovoltaic power generation.

[0092] Equipment operating costs are defined as the amortized additional costs (excluding electricity costs) arising from the depreciation of the equipment's own investment. The calculation parameters are as follows: Standardized cost of electricity (SCO) for photovoltaic power generation is 0.35 yuan / kWh; for wind power generation, it is 0.3 yuan / kWh; for battery cycle costs, it is 0.3 yuan / kWh; for LED supplemental lighting, it is 0.4 yuan / kWh; for electric heating devices, it is 0.15 yuan / kWh; and for electric heating CO2 generators, it is 0.25 yuan / kWh.

[0093] Equipment maintenance costs are calculated based on the rated power of the equipment: 50 yuan / kW per year for photovoltaic modules; 100 yuan / kW per year for wind turbines; 15 yuan / kWh per year for energy storage batteries; and 60 yuan / kW per year for environmental control equipment. For short-term dispatching lasting 1 hour, the annual maintenance cost should be divided by 8760 (total number of hours per year).

[0094] 1.4 Determine constraint parameters

[0095] Power balance constraint: Wind and solar power output in a microgrid + energy storage discharge power + purchased power = load power (including environmental regulation equipment load) and control auxiliary system load + Energy storage charging power + Electricity sales power.

[0096] Microgrid operating constraints: Maximum output power of photovoltaic modules (Installed capacity 50kWp, more accurate predicted value) Maximum output power of wind turbine (Installed capacity 20kW, more accurate predicted value) The energy storage device uses a lithium battery pack with a capacity of 100kWh and a maximum charge / discharge power of [missing information]. The State of Charge (SOC) is constrained within the range of [20%, 80%], with an initial SOC of 50%. Maximum power purchase capacity from the external power grid. Maximum power sales .

[0097] Crop growth environment constraints (determined according to the Technical Specifications for Safe Control of Vegetable Diseases and Pests (GB / T 23416-2009)): light intensity I∈[300,800]μmol / m² / s, temperature T∈[10,35]℃ (minimum nighttime temperature not lower than 10℃, maximum daytime temperature not higher than 35℃), humidity H∈[60,80]%, CO2 concentration C∈[800,1200]ppm.

[0098] 1.5 Definition of Decision Variables

[0099] The decision variables are:

[0100]

[0101] in:

[0102] For continuous power variables;

[0103] Set point variables for environmental factors such as light intensity, temperature, carbon dioxide concentration, and humidity;

[0104] A binary variable representing the on / off mode of device operation.

[0105] S2: Deployment of multi-source data acquisition systems and data prediction

[0106] 2.1 Hardware Deployment of Data Acquisition System

[0107] Image acquisition unit: One Hikvision HD network camera is installed at the front, middle, and rear of the greenhouse, with an 8mm lens focal length and a sampling frequency of once per hour, covering the entire planting area. The cameras are connected to an edge computing gateway (Raspberry Pi 4B) via a PoE switch and have a built-in YOLOv8-based crop feature recognition model for real-time image processing.

[0108] Distributed Sensor Network Unit: A ZigBee wireless sensor network is used to collect spatial distribution data of environmental parameters. Ten monitoring nodes (numbered 1–10) are deployed in a 10m × 10m grid within the greenhouse. Each node integrates: a light intensity sensor, a temperature and humidity sensor (temperature measurement range -40-125℃, accuracy ±0.3℃; humidity measurement range 0-100%RH, accuracy ±2%RH), and a CO2 sensor (measurement range 0-5000ppm, accuracy ±50ppm). Sensor data is transmitted to the data center via a ZigBee gateway (based on an STM32F407 microcontroller), with a sampling frequency of once every 10 minutes.

[0109] Microgrid data monitoring unit: Current sensors, voltage sensors, and power sensors are installed at the photovoltaic inverter, wind power controller, energy storage BMS, and grid connection switch to collect real-time data such as wind and solar power output, energy storage charging and discharging status, load power, and power purchased and sold. The data is transmitted to the microgrid monitoring system (a SCADA system based on an industrial computer) via industrial Ethernet (Profinet protocol) at a sampling frequency of once per second.

[0110] 2.2 Data Acquisition Process

[0111] Taking a specific day during the peak photosynthetic demand period of tomato fruit expansion in the later stages of the experiment as an example, the data collection and preprocessing process is as follows:

[0112] (1) Crop growth status data: Tomato images were collected once per hour from 6:00 to 18:00. The image processing module used the YOLOv8 model to identify the number of fruits per tomato plant, fruit diameter, and leaf color. The average value was used to determine the growth stage as the fruit enlargement stage, and the leaf area index (LAI) was calculated. The average value of the 12 sets of collected image data was calculated, and 2 sets of abnormal data caused by light reflection were removed.

[0113] (2) Environmental parameter data: Data was collected once every 10 minutes from 10 sensor nodes. The average environmental parameters of the greenhouse were calculated by weighted average method (weighted by planting density); abnormal temperature data were removed by 3σ criterion.

[0114] (3) Microgrid operation data: The data sampled every second is averaged over a period of 1 minute to ensure data stability. For example, at t=8:00, the output power of the photovoltaic inverter... (Irradiance 650W / m²), wind power controller output power (Wind speed 3.2m / s), energy storage BMS displays SOC=60%, current charging / discharging power 0kW, load power (Including 8kW supplemental lighting, 2kW ventilation fans, 3kW CO2 generator, 2kW irrigation pump, and 3kW other loads). There is no electricity purchase or sale with the external power grid. , ).

[0115] 2.3 Predictive Data Generation and Model Predictive Control (MPC) Framework

[0116] This embodiment employs a Model Predictive Control (MPC) framework to achieve closed-loop optimization. MPC is an advanced control strategy based on models, rolling optimization, and feedback correction. Its core is that at each decision time, based on the current system state and the predictive model, a future optimization problem in a finite time domain is solved, but only the first control action in the optimal solution sequence is executed; at the next time step, prediction and optimization are performed again based on the new measured state, and so on, rolling forward.

[0117] The prediction model is used to generate hourly data sequences for a future scheduling period M (24 hours), serving as known inputs to the optimization problem. Specifically, at time k, the following prediction model is run to generate predicted sequences for the next M-1 time periods, which serve as known inputs to the optimization problem. :

[0118] Wind and solar power output prediction model: Based on historical wind and solar power output data, real-time meteorological monitoring data, and short-term numerical weather prediction (NWP) for the next 24 hours, it is trained using time series analysis (such as ARIMA) or machine learning algorithms (such as support vector machine regression (SVR) and long short-term memory network (LSTM), and outputs a sequence. This sequence is directly used as the upper limit for the power output constraint of wind and solar power generation.

[0119] Greenhouse Environment Prediction Model: Based on historical data training, the model is a data-driven model that uses outdoor weather forecasts, current indoor environmental conditions, and control commands (such as supplemental lighting and ventilation power) as inputs to predict the spatiotemporal changes of light intensity (I), temperature (T), humidity (H), and CO2 concentration (C) in the greenhouse in the future.

[0120] Load demand forecasting model: Based on environmental targets, calculate the base load of environmental control equipment, and combine historical electricity consumption patterns to predict the time-series demand of other auxiliary loads.

[0121] Each optimization solution follows these steps:

[0122] (1) Prediction: at time t K Using the aforementioned prediction model, and based on the latest measured data collected by S2, a prediction from t is generated. K to t K+M-1 All predicted sequences (M=24 hours).

[0123] (2) Optimization: Using the predicted sequence as known conditions, solve for the time interval [t] defined by S1. K , t K+M-1 The multi-objective optimization problem on ] yields a set of optimal control command sequences U*= .

[0124] (3) Execution: Only execute the first control instruction in the sequence. The data is then distributed to the execution layer (microgrid and environmental control equipment).

[0125] (4) Rolling: Waiting for a scheduling interval Δt (1 hour) to arrive at time t K+1 Repeat steps 1-3. Based on t K+1 The new state at each moment is re-predicted and optimized to achieve closed-loop feedback and overcome prediction errors and unknown disturbances.

[0126] S3: Online optimization solution process based on improved NSGA-III

[0127] Based on real-time acquired data, the multi-objective optimization model constructed in S1 is solved. Its core is an improved NSGA-III algorithm that integrates photosynthetic efficiency priority with grid operation stability considerations. The algorithm parameters are set as follows: population size N=100, maximum number of iterations G. max =600. The specific solution process is as follows:

[0128] 3.1 Encoding of Decision Variables and Population Initialization

[0129] The decision variables to be optimized are encoded into chromosomes using real numbers. Randomly generate an initial population that satisfies all constraints within the feasible region. }

[0130] 3.2 Dynamic parameter adjustment mechanism based on photosynthetic efficiency priority

[0131] The algorithm parameters are not fixed but dynamically adjusted according to the crop's photosynthetic needs. A "photosynthetic efficiency priority coefficient" β is defined, whose value is positively correlated with the current crop growth stage (e.g., β=1.0 during flowering and fruit setting, β=0.8 during vegetative growth) and the current time period (e.g., β is higher during periods of effective photosynthetic radiation). The algorithm's crossover probability Pc and mutation probability Pm are adjusted based on β.

[0132] Calculate the photosynthetic efficiency priority factor for the g-th generation population:

[0133] , among which and In the current population The maximum and minimum values;

[0134] The crossover probability P of the dynamic adjustment algorithm c With the probability of mutation P m The strategy is adjusted as follows:

[0135] , ;

[0136] in, =0.85, =0.08, =0.01; Adjustment constant =0.25, =0.1, =0.65. When > When tanh output is positive, it significantly improves P. c Appropriately reduce P m During the fruit enlargement period, the search and inheritance of high photosynthetic performance scheduling schemes should be strengthened.

[0137] 3.3 Fitness Function Evaluation with Incorporation of Stability Constraints

[0138] For each individual in the population (i.e., a set of scheduling and control schemes), firstly, based on the model of S1, calculate its objective function value F1 within the scheduling period (e.g., the next 24 hours, with 1-hour intervals). (cumulative photosynthetic products) and F2 ( (Net energy cost).

[0139] To assess the impact of the proposed scheme on the operational stability of the microgrid, a stability penalty factor α is defined. The fluctuations in power of the energy storage devices and the power of the interconnection line with the main grid between adjacent time periods are calculated under this scheme. Set an allowable fluctuation threshold. (e.g., 2kW). If all time periods All less than If the value exceeds the limit, then α = 1; if the value exceeds the limit, then α is calculated according to the formula. Decreasing, where η is the attenuation coefficient. This represents the fluctuation amount exceeding the threshold. α∈(0,1], the larger the fluctuation, the smaller the value of α.

[0140] Finally, the fitness value of each individual is calculated. By negatively minimizing the objective F2 and using stability as a multiplicative penalty term, priority selection of the "high photosynthetic output, low energy cost, and stable operation" three-in-one scheme was achieved.

[0141] 3.4 Iterative Evolution

[0142] Selection: Clustering is performed based on non-dominated ordination and dynamic reference points, employing an elite-preserving tournament selection (scale 3). Individuals are non-dominated ordained according to their F1 and -F2 values, dividing them into different frontier layers (Pareto ranks). A set of uniformly distributed reference points is pre-defined in the target space (F1, -F2). Population individuals are associated with the nearest reference point. When generating the next generation population through selection operations, individuals in lower frontier layers are preferentially retained; within the same frontier layer, individuals associated with reference points having fewer associated individuals are preferred to maintain population diversity.

[0143] For continuous variables, simulated binary crossover (SBX) is used, with the distribution index η. c Adapts to the range of variable values; uniform crossover is used for discrete variables with a probability of . .

[0144] For continuous variables, multinomial variation (PM) is used, with the distribution index η. m =25, applying bit-flip mutation to the discrete variable, with a probability of... .

[0145] 3.5 Iteration Termination and Optimal Strategy Output

[0146] Repeat steps 3.2 to 3.4 until the preset maximum number of iterations G is reached. max Or it satisfies the hypervolume convergence criterion The reference point for the hypervolume (HV) index is set as follows: , For the target point that is currently close to the forefront, the convergence threshold is... =0.0005.

[0147] The first non-dominated frontier (Pareto optimal solution set) is extracted from the final generation population. Decision-makers can select the most suitable compromise solution from this set by adjusting the weighting coefficients ω1 and ω2 according to current management preferences. Decoding this selected solution yields the microgrid dynamic scheduling instructions for each time period within a future scheduling cycle. .

[0148] S4: Control Implementation

[0149] The optimization solution module sends the first instruction (corresponding to the current time) in the joint optimization strategy sequence via the communication network. After the instruction for the current time period is executed, the system waits for a 1-hour interval. Upon reaching the next scheduling time, a new round of data acquisition (S2) and optimization (S3) is immediately triggered, and a new first instruction is executed. This process is repeated cyclically to achieve dynamic adaptation. Specifically, this includes:

[0150] The execution control layer parses and executes the first instruction in the joint optimization policy sequence. The execution of microgrid dynamic dispatch commands (through the microgrid control unit) includes issuing active power setpoints to photovoltaic inverters and wind turbine converters. , Send charging and discharging commands to the energy storage converter. , Control the static switch (STS) at the grid connection point to perform electricity purchase and sale operations. The execution of greenhouse environment coordinated control commands (via the environmental control unit) includes adjusting the intensity of LED supplemental lighting via the dimming system. Set the target temperature of the variable frequency air conditioning system via Modbus communication. The system controls the opening of ventilation windows and the speed of the circulating fan to assist in temperature and humidity regulation; the PID controller adjusts the valve opening of the CO2 generator to bring the concentration towards a suitable level. The humidity is adjusted by the humidifier controller to tend towards... .

[0151] Rolling optimization. Wait for a scheduling interval Δt = 1 hour, and the system clock advances to time k+1. Immediately return to step S2, and based on the latest acquired real-time status data k+1 and the updated prediction sequence, repeat the optimization solution in step S3 to obtain a new sequence. and execute its first instruction again. This process is repeated continuously to achieve closed-loop feedback and rolling optimization, effectively overcoming prediction errors and unknown disturbances.

[0152] Example 2: Hardware Implementation of a Greenhouse Microgrid Dynamic Scheduling and Photosynthetic Efficiency Co-optimization Control System

[0153] This embodiment provides a specific hardware implementation scheme for a greenhouse microgrid dynamic scheduling and photosynthetic efficiency coordinated optimization control system. Corresponding to the method in Embodiment 1, the system adopts a four-layer architecture of "perception layer - network layer - decision layer - execution layer". The hardware selection and connection relationship of each layer are as follows:

[0154] Perception layer (data acquisition module hardware)

[0155] (1) Image acquisition unit: 3 Hikvision HD network cameras (4 million pixels, infrared night vision function, protection level IP67), installed at a height of 3m, with horizontal angles of 0°, 120° and 240° respectively; the edge computing gateway is a Raspberry Pi 4b with a built-in YOLOv8 model inference engine.

[0156] (2) Distributed sensor network unit: 10 customized sensor nodes (core chip STM32F103), each node integrates: light sensor, temperature and humidity sensor, CO2 sensor; customized Zigbee gateway (core chip STM32F407), the Zigbee chip is CC2530; the sensor power supply adopts solar power (5W photovoltaic panel + 10Ah lithium battery) to ensure continuous operation.

[0157] (3) Microgrid data monitoring unit: current sensor measurement range 0-50A, accuracy ±1%, voltage sensor measurement range 0-1000V, accuracy ±1%, power sensor measurement accuracy ±0.1%; the data acquisition card is selected with 16 channels and 16-bit AD conversion, and is installed on the microgrid monitoring host.

[0158] Network layer

[0159] The system adopts a multi-network converged architecture of "Industrial Ethernet + Zigbee + 5G": between the perception layer and the decision layer, microgrid data is transmitted via Profinet Industrial Ethernet (transmission rate 100Mbps, latency ≤1ms), while environmental data and crop image data are transmitted via Zigbee (transmission rate 125kbps, latency ≤100ms); between the decision layer and the remote monitoring center, data is transmitted via a 5G network, supporting remote data viewing and command issuance.

[0160] Decision-making layer (hardware for model building and optimization solution module)

[0161] The system uses an industrial control computer (IPC) as the core decision-making unit, specifically the Advantech IPC-610L (equipped with an Intel Core i7-12700 processor, 32GB of memory, and TB of SSD storage). It includes a built-in model building program (developed based on MATLAB R2023a) and an improved NSGA-III algorithm solver (developed based on Python 3.9+DEAP framework). It is also equipped with a 15-inch industrial touchscreen that supports local parameter setting and status monitoring.

[0162] Execution layer (execution control module hardware)

[0163] (1) Microgrid control unit: The photovoltaic controller is Huawei SmartLogger 3000 (supports MPPT regulation, maximum access power 60kW), the wind power controller is a 20kW wind turbine dedicated controller (supports pitch control), the energy storage charge and discharge controller is CATL BMS-100kWh (supports charge and discharge protection and SOC estimation), and the grid connection switch is Schneider or Chint NW series molded case circuit breaker with rated current of 250A, with electric operating mechanism and communication module (supports Modbus); each controller communicates with the decision-making level through Modbus-RTU protocol.

[0164] (2) Environmental control unit: The customized mode of STM32F407 control board + interface driver module is adopted. For example, the supplementary light uses a high-power MOSFET / solid-state relay module; the shading control uses a stepper motor driver; the ventilation control uses a single-phase fan speed control module or a small frequency converter; the CO2 control combines analog high-precision PID control of concentration; the heating control uses a thyristor power regulator or a multi-channel AC contactor; and the humidification control uses a solid-state relay.

[0165] Data storage and interaction module

[0166] It adopts a MySQL distributed database (a cluster consisting of 2 servers with a storage capacity of 10TB) to store the collected raw data, model parameters, scheduling instructions, and control results; it is equipped with a storage array to support data backup and disaster recovery; and it communicates with the monitoring platform (based on WebGIS) of the remote monitoring center (deployed in the agricultural technology extension center) through a 5G module to realize real-time data uploading, remote monitoring, and instruction issuance.

[0167] System power supply and protection

[0168] The system power supply adopts a dual power supply mode of "microgrid + backup mains power" and ensures stable power supply through ATS automatic switching switch; all outdoor equipment (cameras, sensors, gateways) adopts IP67 protection level, and the control cabinet adopts IP54 protection level. It is equipped with dehumidifier and surge protector to adapt to the high temperature and high humidity environment of greenhouse.

[0169] Through the above specific implementation methods, those skilled in the art can construct and run this system. In a complete simulation of a tomato fruit expansion period (approximately 30 days), compared with the traditional "lowest energy cost" scheduling strategy, the method of this invention is expected to achieve the following: under the premise that the net energy cost of the microgrid is basically the same or slightly reduced (<6%), the cumulative net photosynthetic products of tomatoes will increase by 8%-17%, laying the foundation for the accumulation of dry matter in the fruit; the wind and solar curtailment rate will be reduced by more than 20%, and surplus electricity will be actively used to create "photosynthetic opportunity windows" (such as supplementing light in advance during off-peak electricity prices and increasing CO2 application); thanks to the stability penalty factor, the daily average power change rate of key power equipment (such as energy storage converters) will be reduced by more than 20%, and the equipment lifespan will be extended.

[0170] The scope of protection of this invention is not limited to the above embodiments. For different crops (such as cucumbers and lettuce), adaptation can be achieved by adjusting the photosynthetic rate model parameters and environmental constraint thresholds. For greenhouses of different sizes, the microgrid capacity parameters and the number of sensors deployed can be adjusted. The collaborative optimization intelligent algorithm can also be replaced with other intelligent algorithms with multi-objective optimization capabilities, such as improved particle swarm optimization algorithms, improved genetic algorithms, etc. All solutions that conform to the core ideas of this invention should be within the scope of protection of this invention.

Claims

1. A method for dynamic scheduling and synergistic optimization control of photosynthetic efficiency in greenhouse microgrids, characterized in that, Includes the following steps: S1: Construct a multi-objective energy management and scheduling model with the goal of maximizing the cumulative net photosynthetic products of crops and minimizing the net energy cost of the microgrid within the scheduling period M, while simultaneously satisfying the microgrid operation constraints and crop growth environment constraints. S2: Real-time acquisition of system status data This includes spatial distribution data of greenhouse internal environmental parameters, crop growth status data, and distributed power generation output, energy storage status of charge, load demand, and grid-connected power purchase and sale data in the microgrid; and prediction sequences of environmental parameters, wind and solar power generation output, and load demand within the future scheduling period [k+1, k+M-1] generated by the prediction model. S3: Real-time data collected in step S2 and predicted sequence A multi-objective optimization algorithm is used to solve the model constructed in step S1 online, and the result is a joint optimization strategy sequence that includes both microgrid dynamic scheduling instructions and greenhouse environment coordinated control instructions for the next scheduling cycle. S4: Execute the control command corresponding to the current time in the joint optimization strategy sequence. After a preset scheduling interval, at time k+1, the process returns to step S2, regenerates the prediction sequence based on the latest collected real-time data, repeats the optimization solution in step S3, and executes the process in step S4, thereby achieving closed-loop feedback and rolling optimization.

2. The method as described in claim 1, characterized in that, In step S1, the objective function of the multi-objective energy management and scheduling model is constructed as follows: Objective function F = ω1 × k yp ×F1+ω2×F2, where F1 is the objective function for maximizing cumulative photosynthetic products, F2 is the objective function for minimizing net energy costs, and ω1 and ω2 are weighting coefficients that satisfy ω1+ω2=1; F1 is constructed based on a crop photosynthetic rate model, and its expression is: In the formula, M is the scheduling cycle, S is the greenhouse planting area, and P is the greenhouse planting area. n The net photosynthetic rate of the crop is given by the Farquhar model, where I(t,s) is the light intensity at time t and location s inside the greenhouse, T(t,s) is the temperature at time t and location s inside the greenhouse, C(t,s) is the carbon dioxide concentration at time t and location s inside the greenhouse, H(t,s) is the humidity at time t and location s inside the greenhouse, and k is the net photosynthetic rate of the crop. yp To convert the dimensions of F1 and F2 to be consistent, the expression for F2 is: in: 、 These represent the electricity purchased and sold by the microgrid to the external power grid, respectively. 、 、 、 These represent the power generation capacity of photovoltaic and wind power, and the charging and discharging power of energy storage batteries, respectively. 、 、 、 These are the unit operating costs of photovoltaic, wind power, energy storage batteries, and environmental equipment, respectively. 、 These represent the unit costs of purchasing and selling electricity from the external power grid to the microgrid, respectively. Power consumption of environmental equipment, such as supplemental lighting, heating devices, and CO generators; The objective function's decision variables are: the power generation capacity of wind and solar power generation equipment and the charging and discharging power of energy storage batteries. ,in, For continuous power variables, Set point variables for the environment. This is a binary variable representing the on / off mode of equipment operation. The constraints of the multi-objective energy management and scheduling model include microgrid operation constraints and crop growth environment constraints; the microgrid operation constraints include at least power balance constraints, wind and solar power output constraints, energy storage device charging and discharging constraints, and electricity purchase and sale constraints; wherein, the power balance constraint is: The power output constraints for wind and solar power generation are: ， The charging and discharging constraints of energy storage devices are: The constraints on the purchase and sale of electricity are: ， The constraints of the crop growth environment are as follows: In the formula, I min,i I max,i These represent the minimum and maximum suitable light intensity during the growth period of crop i, respectively, T. min,i T max,i These represent the minimum and maximum suitable temperatures during the growth period of crop i, H. min,i H max,i These represent the minimum and maximum suitable humidity values ​​during the growth period of crop i, respectively. min,i C max,i These represent the minimum and maximum suitable carbon dioxide concentrations during the growth period of crop i, respectively.

3. The method as described in claim 1, characterized in that, The multi-objective optimization algorithm mentioned in step S3 is the Non-Dominated Sorting Genetic Algorithm (NSGA-III). The improvement of this algorithm is as follows: the crossover probability and mutation probability are dynamically adjusted based on the priority of crop photosynthetic efficiency, and a microgrid operation stability constraint factor is introduced to optimize the fitness function; its online solution process specifically includes: S31: Population initialization: N sets of initial solutions that satisfy all constraints are randomly generated within the feasible region. Each solution set includes microgrid scheduling variables and environmental control variables; the microgrid scheduling variables include wind and solar power generation output allocation coefficients, energy storage device charging and discharging power, and power purchased and sold from the external grid; the environmental control variables include the setting parameters of supplemental lighting, temperature control devices, carbon dioxide generators, and humidifiers; S32: Dynamic parameter adjustment: for the g-th generation population P g Based on the current crop growth period photosynthetic efficiency priority factor β g The crossover probability P of the dynamic adjustment algorithm c With the probability of mutation P m The strategy is adjusted as follows: ， ;in, ， 、 and To adjust the constant and ensure that P is appropriately increased during the peak photosynthetic demand period of crops. c And reduce P m S33: Fitness Assessment: Calculate the fitness of each individual in the current population. fitness value Where α is the microgrid operation stability constraint factor, and its value is related to the power fluctuation between adjacent scheduling periods. Negative correlation, the strategy is: ,in This indicates that the average value over the scheduling period is used to penalize scheduling schemes that cause drastic power fluctuations; S34: Iterative evolution: The population is clustered based on non-dominated sorting and reference vectors, and tournament selection is used; simulated binary crossover (SBX) is used for continuous variables, and uniform crossover is used for discrete variables, with a probability of... For continuous variables, polynomial mutation (PM) is used; for discrete variables, bit-flip mutation is used, with a probability of... S35: Convergence Judgment and Output: Repeat steps S32 to S34 until g ≥ G is satisfied. max (Maximum number of iterations) or (HV is the hypervolume index) The process terminates when the population reaches a certain threshold. From the non-dominated solution set of the final generation, a compromise solution chr is selected based on linear scalarization using weight coefficients ω1 and ω2 according to a preset preference. ∗ The decoded output is a joint optimization strategy U that includes specific power values ​​and environmental device settings. ∗ .

4. The method as described in claim 1, characterized in that, In step S2, spatial distribution data of greenhouse environmental parameters are collected through a distributed wireless sensor network; crop types, growth stages, and leaf area index are identified through a machine vision system; real-time power data are collected through a microgrid energy management system; the prediction model includes a wind and solar power output prediction model, a greenhouse environment prediction model, and a load demand prediction model.

5. The method as described in claim 2, characterized in that, The microgrid dynamic scheduling commands are based on decision variables of the objective function, including the output setpoints of photovoltaic inverters and wind turbine converters, the charging and discharging power commands of energy storage converters, and the power exchange commands at the grid connection point connected to the external power grid; the environmental coordinated control commands include the setpoints of supplementary lighting intensity, air conditioning system set temperature, CO2 generator rate, and humidifier power.

6. A cooperative optimization control system for implementing the method of any one of claims 1-5, characterized in that, The system has a hierarchical structure, including: a perception and prediction layer: consisting of distributed IoT nodes, machine vision units, and state estimators, responsible for implementing the S2 function; an optimization decision layer: consisting of high-performance embedded computing units running improved evolutionary algorithms, responsible for implementing the S3 function; and an execution control layer: consisting of programmable logic controllers and power converter drive circuits, responsible for implementing the S4 function. The layers interact with each other through a real-time data bus to form a closed loop.

7. The system as described in claim 6, characterized in that, The data acquisition and prediction module includes an image acquisition unit, a distributed sensor network unit, a microgrid data monitoring unit, and a data processing server connected to the above units for running prediction models. The image acquisition unit includes a high-definition camera and an image processing module, wherein the high-definition camera is used to acquire crop images, and the image processing module is used to analyze the crop images to obtain crop growth status data. The distributed sensor network unit includes sensor nodes and gateways evenly deployed in different areas of the greenhouse. The sensor nodes integrate various sensors for acquiring environmental parameter data of different areas of the greenhouse. The microgrid data monitoring unit is used to collect real-time data on the output of wind and solar power generation equipment, the charging and discharging status of energy storage equipment, load power, and purchased and sold electricity.

8. The system as described in claim 6, characterized in that, The execution control module includes a microgrid control unit and an environmental control unit; the microgrid regulation unit is connected to the wind and solar power generation equipment, energy storage equipment, and grid connection switch in the microgrid, and is used to execute the microgrid dynamic scheduling command; the environmental regulation unit is connected to the supplementary lighting, temperature control device, ventilation device, CO2 generator, and humidifier in the greenhouse, and is used to execute the greenhouse environment coordinated regulation command.

Images