Intelligently regulated and controlled photovoltaic energy storage micro-grid cooperative operation system and intelligent regulated and controlled photovoltaic energy storage micro-grid cooperative operation method

Abstract

The invention relates to an intelligent regulation and control photovoltaic energy storage micro-grid cooperative operation system and an intelligent regulation and control photovoltaic energy storage micro-grid cooperative operation method. Illumination and load space-time changes are accurately grasped through a space-time big data prediction technology, energy flow intelligent regulation and control strategies are utilized to realize energy optimization distribution, equipment operation efficiency is improved by means of a multi-equipment collaborative optimization mechanism, economy and low carbon are considered based on a low-carbon economic dual-objective optimization operation model, and the method is suitable for large-scale popularization and application. And the power supply reliability is guaranteed by an adaptive fault diagnosis and rapid recovery system. The system collects multi-source data, and efficient, stable and low-carbon operation of the micro-grid is achieved through the steps of prediction, regulation, coordination, optimization, diagnosis and the like. Practical cases show that the technology can reduce the light abandoning rate by 12%, reduce the power purchasing cost of the main network by 40%, reduce the carbon emission by 20%, and significantly improve the comprehensive performance and sustainable development capability of the micro-grid.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of new energy power generation and smart grid, in particular to a kind of intelligent control's photovoltaic energy storage micro-grid collaborative operation system and method, to realize the efficient, stable, low carbon operation of photovoltaic energy storage micro-grid, improve renewable energy consumption capacity and energy comprehensive utilization efficiency. BACKGROUND

[0002] (1) Problems existing in the prior art

[0003] Poor spatiotemporal matching: traditional photovoltaic energy storage micro-grid has insufficient adaptability to the spatiotemporal changes of light and load. Influenced by day-night alternation and weather changes, photovoltaic power generation has obvious spatiotemporal fluctuations, and user load demand is also unevenly distributed over time and space. The existing system cannot accurately predict power generation and power consumption in different regions and time periods, leading to energy supply and demand imbalance, frequent light abandonment, and low energy utilization rate.

[0004] Extensive energy flow regulation: the energy flow of photovoltaic, energy storage, load, and interaction with the main grid in the micro-grid lacks fine regulation. The charging and discharging strategy of the energy storage system is single, and problems such as overcharging or insufficient discharging often occur; photovoltaic output and load power consumption cannot be matched in real time, and excess power cannot be effectively utilized, and when insufficient, it relies on high-cost main grid power purchase, increasing operating costs and reducing energy self-sufficiency.

[0005] Insufficient multi-device collaboration: the photovoltaic array, energy storage device, converter, transformer, and other devices in the micro-grid have low collaborative operation efficiency. The control of each device is relatively independent, lacking a unified coordination mechanism, leading to unreasonable power distribution between devices, overloading of some devices, shortening of device life, and affecting the overall stability and reliability of the micro-grid.

[0006] It is difficult to achieve the goal of low-carbon operation: in the existing micro-grid operation and management, there is a lack of effective means to deeply integrate the low-carbon goal into the operation strategy. It is difficult to accurately quantify carbon emissions under different operation modes, making it difficult to develop an optimization scheme that takes into account both economy and low carbon, which is not conducive to achieving the "double carbon" goal and limiting the sustainable development of micro-grids. SUMMARY

[0007] The present application aims to provide a photovoltaic energy storage micro-grid collaborative operation system based on spatiotemporal dynamic optimization and energy flow intelligent regulation, comprising:

[0008] A spatiotemporal big data prediction module for collecting light, weather, and load data, using a spatiotemporal convolutional neural network and a long short-term memory network fusion model to predict photovoltaic power generation and load demand; an energy flow intelligent regulation module for establishing a micro-grid energy flow mathematical model and implementing rolling optimization distribution of energy flow using a model predictive control algorithm.

[0009] A multi-device cooperative optimization module realizes the cooperative operation of devices in the micro-grid through a distributed cooperative control architecture based on a consensus algorithm and a distributed optimization algorithm.

[0010] A low-carbon economy double-target optimization module constructs a double-target optimization operation model containing economic cost and carbon cost, adopts a multi-objective particle swarm optimization algorithm to solve and select the optimal operation scheme.

[0011] An adaptive fault diagnosis and rapid recovery module uses intelligent sensors to collect data, performs fault diagnosis through a convolutional neural network, and realizes fault isolation and rapid system recovery.

[0012] Further, the prediction model of the spatio-temporal big data prediction module is trained by historical data and can update the prediction of power generation and load data in different regions within 15 minutes to 24 hours in real time.

[0013] Further, the energy flow intelligent regulation and control module minimizes economic cost and carbon emissions as the goal, and performs rolling optimization on the energy storage charge and discharge power, photovoltaic output power and main grid interaction power in multiple time periods in the future.

[0014] Further, in the multi-device cooperative optimization module, each device interacts with the real-time operation state and power demand information through a unified data interface protocol to realize cooperative decision-making.

[0015] Further, the low-carbon economy double-target optimization module introduces a carbon trading mechanism and a carbon emission factor, and adopts a fuzzy decision method to select the optimal scheme from the Pareto frontier solution set.

[0016] Further, the convolutional neural network fault diagnosis model of the adaptive fault diagnosis and rapid recovery module can quickly identify fault types and locations, and automatically perform fault isolation and recovery operations.

[0017] Further, the photovoltaic energy storage micro-grid cooperative operation method is characterized by comprising the following steps:

[0018] Collecting light, weather, and load data, using a spatio-temporal convolutional neural network and a long short-term memory network fusion model to predict photovoltaic power generation and load demand;

[0019] Establishing a micro-grid energy flow mathematical model, taking the minimization of economic cost and carbon emissions as the goal, and using a model predictive control algorithm to perform rolling optimization and distribution of energy flow;

[0020] Through a distributed cooperative control architecture, based on a consensus algorithm and a distributed optimization algorithm, the cooperative operation of devices in the micro-grid is realized.

[0021] A dual-objective optimization operation model incorporating economic and carbon costs is constructed, which is then solved using a multi-objective particle swarm optimization algorithm, and the optimal operation scheme is selected through fuzzy decision-making.

[0022] By using smart sensors to collect data and using convolutional neural networks for fault diagnosis, fault isolation and rapid system recovery can be achieved.

[0023] Furthermore, in the prediction step, the model is trained using historical data and updated every 15 minutes with prediction data for the next 15 minutes to 24 hours.

[0024] Furthermore, in the energy flow regulation step, the energy allocation scheme for the next 4 hours is optimized in 15-minute intervals, and the control command for the current time period is executed.

[0025] Furthermore, in the fault diagnosis and recovery steps, after the convolutional neural network model identifies the fault, the system completes fault isolation and power restoration operations within 5 minutes.

[0026] The present invention has the following beneficial effects:

[0027] This invention proposes a precise prediction technology based on spatiotemporal big data, constructing a spatiotemporal database containing historical sunshine, meteorological, and load data. A fusion model of spatiotemporal convolutional neural network (STCNN) and long short-term memory network (LSTM) is used to predict photovoltaic power generation and load demand in different regions and time periods. By extracting time-series features and spatial correlation features, the spatiotemporal variation patterns of sunshine and load are accurately captured, providing a reliable data foundation for the optimized operation of microgrids.

[0028] To address the problem of inefficient energy flow regulation, an intelligent energy flow regulation strategy is designed. A mathematical model of microgrid energy flow is established, comprehensively considering factors such as photovoltaic output, energy storage charging and discharging, load demand, and power interaction with the main grid. A model predictive control (MPC) algorithm is employed, with the optimization objective of minimizing economic cost and carbon emissions, to continuously optimize the energy flow allocation scheme for multiple future time periods. Based on real-time forecast data and system operating status, the energy storage charging and discharging strategy, photovoltaic power output, and main grid power purchase and sale plan are dynamically adjusted to achieve efficient energy utilization and optimized allocation.

[0029] This invention establishes a multi-device collaborative optimization mechanism, treating photovoltaic (PV) arrays, energy storage systems, and converters within a microgrid as collaborative units. Through a distributed collaborative control architecture, each device exchanges information such as operating status and power demand in real time. Based on consensus and distributed optimization algorithms, it achieves rational power allocation and collaborative operation among devices. For example, when sunlight is abundant, the PV array, energy storage system, and load devices are coordinated to prioritize meeting local load demands, with excess energy stored in the energy storage system. During peak load periods and when PV power is insufficient, the energy storage system discharges, and the main grid purchases power rationally, ensuring stable system operation.

[0030] To achieve the goal of low-carbon operation, a dual-objective optimization model for low-carbon economy is constructed. A carbon trading mechanism and carbon emission factors are introduced to quantify carbon emissions during microgrid operation. Carbon costs are incorporated into the objective function, forming a dual-objective optimization system together with economic costs. The model is solved using the multi-objective particle swarm optimization algorithm (MOPSO) to obtain the Pareto front solution set. A fuzzy decision-making method is then used to select the optimal operating scheme, ensuring economical operation of the microgrid while minimizing carbon emissions.

[0031] To address potential equipment failures and operational anomalies in microgrids, an adaptive fault diagnosis and rapid recovery system was developed. Smart sensors are deployed at key nodes of the microgrid to collect real-time operating parameters such as voltage, current, and temperature. A fault diagnosis model is constructed using convolutional neural networks (CNNs) from deep learning to quickly and accurately identify the type and location of faults. When a fault occurs, the system automatically isolates the faulty area, activates backup equipment, or adjusts operating strategies to achieve rapid recovery of the microgrid and ensure power supply reliability. Attached Figure Description

[0032] Figure 1 Overall process flow diagram. Detailed Implementation

[0033] The embodiments of the present invention will be described in further detail below with reference to the accompanying drawings and examples. The following examples are for illustrative purposes only and should not be construed as limiting the scope of the invention.

[0034] Example 1

[0035] An industrial park plans to construct a 10MWp photovoltaic energy storage microgrid system to meet the electricity needs of businesses within the park, increase the proportion of renewable energy consumption, and reduce carbon emissions. The park contains various types of businesses with significantly different load characteristics, exhibiting marked peak-valley variations. Furthermore, while the area boasts abundant solar resources, frequent weather changes result in strong fluctuations in photovoltaic power generation.

[0036] Under traditional operating modes, microgrids face numerous problems: photovoltaic power generation cannot be effectively absorbed, with an average curtailment rate as high as 18%; unreasonable charging and discharging strategies of energy storage systems lead to shortened battery life and increased maintenance costs; the interaction between microgrids and the main grid lacks optimization, resulting in high electricity purchase costs; in addition, the system lacks an effective fault response mechanism, and once a fault occurs, the power restoration time is long, affecting the normal production of enterprises.

[0037] To address the aforementioned issues, this embodiment introduces a photovoltaic energy storage microgrid collaborative operation system based on spatiotemporal dynamic optimization and intelligent energy flow regulation, with the following objectives:

[0038] Reduce the curtailment rate of photovoltaic power generation to below 5%;

[0039] This will increase the energy self-sufficiency rate of microgrids to over 70%.

[0040] Reduce the overall operating cost of microgrids by 20%;

[0041] The average fault recovery time has been reduced to less than 10 minutes.

[0042] II. Core Technology Implementation Steps

[0043] (I) System Hardware Deployment

[0044] Photovoltaic array installation: Monocrystalline silicon photovoltaic modules will be installed on the rooftops and vacant land within the park, with a total installed capacity of 10MWp. The photovoltaic modules will be installed at an angle optimized to 35° based on the local latitude to maximize sunlight exposure. Each photovoltaic array will be equipped with a smart combiner box and a DC distribution cabinet to enable real-time acquisition and transmission of power generation data.

[0045] Energy storage system configuration: A 6MWh lithium battery energy storage system will be constructed, employing a modular design for easy future expansion. The system will be equipped with a bidirectional converter for AC-DC conversion, with a rated power of 3MW and a conversion efficiency of ≥98%. A battery management system (BMS) will also be deployed to monitor battery voltage, current, temperature, and other parameters in real time, ensuring the safe operation of the energy storage system.

[0046] Sensor network setup:

[0047] Current and voltage sensors were installed in each branch of the photovoltaic array, with accuracies of 0.5% and 0.2%, respectively.

[0048] Temperature sensors (accuracy ±1℃) and current sensors are deployed in key locations such as battery clusters and converters in energy storage systems;

[0049] Smart meters are installed on each feeder branch of the microgrid and at the connection point with the main grid to realize electricity metering and data uploading;

[0050] Install meteorological monitoring stations to collect meteorological data such as light intensity, temperature, humidity, and wind speed in real time.

[0051] Communication and control equipment deployment: A hybrid networking scheme of 5G communication network and fiber optics is adopted to ensure the stability and real-time performance of data transmission. Edge computing devices are deployed to enable local data processing and analysis; a central control server is built to run the microgrid collaborative operation management platform.

[0052] (II) Software System Implementation

[0053] Spatiotemporal Big Data Prediction Module

[0054] Data Acquisition and Storage: Real-time data on illumination, weather, and load are collected via a sensor network at a frequency of once per minute. The data is stored in an InfluxDB time-series database and simultaneously backed up to a MySQL relational database.

[0055] Model Training: Historical data from the past three years was collected to train a fusion model of a spatiotemporal convolutional neural network (STCNN) and a long short-term memory network (LSTM). The model input included historical data on light intensity, temperature, humidity, and load power, and the output was a forecast of photovoltaic power generation and load demand for the next 15 minutes to 24 hours. After repeated parameter tuning and optimization, the root mean square error (RMSE) of photovoltaic power generation prediction was reduced to 8%, and the RMSE of load demand prediction was reduced to 10%.

[0056] Real-time prediction: Input the collected data into the trained model, update the prediction results every 15 minutes, and display the prediction curve through a visualization interface.

[0057] Intelligent Energy Flow Control Module

[0058] Model Establishment: Based on the microgrid topology and equipment parameters, a mathematical model of energy flow is established. Model constraints include:

[0059] Power balance constraints: in Photovoltaic power generation capacity, For charging and discharging of energy storage systems

[0060] Electrical power (negative for charging, positive for discharging) The main grid's power purchase and sales volume (purchased power is positive, sold power is negative). For load power;

[0061] Energy storage system capacity constraints: SOC refers to the state of charge of the energy storage system.

[0062] Equipment power constraints: It represents equipment such as photovoltaics, energy storage, and converters.

[0063] Optimize the target setting; the objective function is: in Main grid electricity purchase cost For the operation and maintenance costs of energy storage systems, Cost of carbon emissions These are weighting coefficients, set to 0.5, 0.3, or 0.2 according to actual needs.

[0064] Rolling optimization: A model predictive control (MPC) algorithm is used to perform rolling optimization of the energy flow for the next 4 hours, with a control cycle of 15 minutes. Each cycle calculates the energy storage charging and discharging power, photovoltaic output power, and grid interaction power commands for the current cycle, and sends them to the corresponding equipment for execution.

[0065] Multi-device collaborative optimization module

[0066] Device interconnection: Each device connects to the communication network via the Modbus TCP / IP protocol to achieve real-time interaction of information such as operating status and power requirements.

[0067] Collaborative decision-making: When the light intensity suddenly increases, the photovoltaic array detects the power increase and sends the information to the central control system. Based on the current load and energy storage status, the system calculates the charging power of the energy storage system and the increase in adjustable load using a consensus algorithm, and then issues instructions to each device to achieve collaborative consumption of excess photovoltaic power.

[0068] Low-carbon economy dual-objective optimization module

[0069] Carbon emission quantification: The carbon emissions during microgrid operation are calculated based on the local grid carbon emission factor (0.6 tCO2 / MWh) and the carbon emission coefficient during the charging and discharging process of the energy storage system. For example, the carbon emissions generated by purchasing 1 MWh of electricity from the main grid are 0.6 tCO2.

[0070] Model solution: The bi-objective optimization model was solved using the multi-objective particle swarm optimization algorithm (MOPSO). After 500 iterations, the Pareto front solution set was obtained.

[0071] Option selection: Using fuzzy decision-making methods, the optimal operating option is selected from the Pareto frontier solution set, taking into account both economic costs and carbon emissions. For example, in a certain period, the option that reduces total costs by 15% and carbon emissions by 12% is selected.

[0072] Adaptive Fault Diagnosis and Fast Recovery Module

[0073] Data acquisition and feature extraction: Intelligent sensors collect data such as voltage, current, and temperature of the equipment in real time, and extract fault feature vectors through methods such as Fast Fourier Transform (FFT).

[0074] Fault diagnosis: Input the feature vector into the trained convolutional neural network (CNN) fault diagnosis model. The model can determine the fault type (such as short circuit, overload, equipment failure, etc.) and fault location within 1 second, with an accuracy of over 95%.

[0075] Fault Handling: When a short-circuit fault is detected in a photovoltaic branch, the system automatically disconnects the circuit breaker of that branch and simultaneously activates the backup power supply to ensure power supply to critical loads. The system also notifies maintenance personnel of the fault location and type via SMS, and maintenance personnel arrive on-site within 10 minutes to carry out repairs.

[0076] III. Implementation Results Verification

[0077] (I) Operational Data Statistics

[0078] After three months of trial operation, the system's operational data is as follows:

[0079]

[0080] (II) Typical Daily Operation Analysis

[0081] A detailed analysis was conducted on a sunny summer day.

[0082] 08:00-12:00: As sunlight intensity gradually increases, photovoltaic power generation rises from 2MW to 8MW. Based on the predicted load demand and energy storage status, the system stores excess electricity in the energy storage system at a capacity of 1.5MW, while simultaneously reducing power purchases from the main grid.

[0083] 12:00-14:00: Sunlight reaches its peak, and photovoltaic power generation remains at around 9MW. At this time, load demand is also at its peak (6MW), the energy storage system continues to charge, and at the same time sells 0.5MW of electricity to the main grid.

[0084] 14:00-18:00: As sunlight intensity decreases, photovoltaic power generation drops to 5MW, while load demand remains high (7MW). The energy storage system discharges at 2MW, and any shortfall is met by purchasing power from the main grid to ensure a stable power supply to the load.

[0085] 18:00-24:00: As night falls, photovoltaic power generation stops, and the energy storage system continues to discharge to meet part of the load demand until the state of charge drops to 20%, at which point the main grid provides full power.

[0086] (III) Economic Benefit Analysis

[0087] Direct economic benefits: It can reduce the amount of electricity purchased from the main grid by about 1.2 million kWh per year, which, calculated at 0.8 yuan / kWh, saves 960,000 yuan in electricity purchase costs; it can also reduce the maintenance costs of the energy storage system by about 150,000 yuan, resulting in a total annual economic benefit increase of 1.11 million yuan.

[0088] Environmental benefits: It can reduce carbon dioxide emissions by approximately 720 tons per year, helping the park achieve its low-carbon development goals.

[0089] IV. Implementation Summary

[0090] This embodiment successfully addresses the problems of insufficient energy absorption, high operating costs, and low reliability inherent in traditional microgrids by deploying a photovoltaic energy storage microgrid collaborative operation system based on spatiotemporal dynamic optimization and intelligent energy flow regulation. The system's modules work collaboratively, achieving accurate prediction of photovoltaic power generation, intelligent regulation of energy flow, efficient equipment coordination, and rapid fault handling. Actual operating data demonstrates that the system has achieved its expected goals, exhibiting significant economic and environmental benefits, and providing a successful and referable example for the optimized operation of photovoltaic energy storage microgrids.

[0091] The embodiments of the present invention are given for illustrative and descriptive purposes only, and are not intended to be exhaustive or to limit the invention to the forms disclosed. Many modifications and variations will be apparent to those skilled in the art. The embodiments were chosen and described in order to better illustrate the principles and practical application of the invention, and to enable those skilled in the art to understand the invention and to design various embodiments with various modifications suitable for a particular purpose.

Claims

1. A smart-controlled photovoltaic energy storage microgrid collaborative operation system, characterized in that, include:

1. Spatiotemporal big data prediction module, used to collect data on sunlight, weather, and load, and use a fusion model of spatiotemporal convolutional neural network and long short-term memory network to predict photovoltaic power generation and load demand; 2. Intelligent energy flow control module: Establishes a mathematical model of microgrid energy flow and uses model predictive control algorithm to achieve rolling optimization allocation of energy flow; 3. Multi-device collaborative optimization module: Through a distributed collaborative control architecture, it realizes the collaborative operation of devices within the microgrid based on consensus algorithms and distributed optimization algorithms; 4. Low-carbon economy dual-objective optimization module: Construct a dual-objective optimization operation model that includes economic costs and carbon costs, and use a multi-objective particle swarm optimization algorithm to solve and select the optimal operation scheme; 5. Adaptive fault diagnosis and rapid recovery module: It uses intelligent sensors to collect data and performs fault diagnosis through convolutional neural networks to achieve fault isolation and rapid system recovery.

2. The system according to claim 1, characterized in that, The prediction model of the spatiotemporal big data prediction module is trained with historical data and can update and predict power generation and load data in different regions in real time for the next 15 minutes to 24 hours.

3. The system according to claim 1, characterized in that, The intelligent energy flow control module aims to minimize economic costs and carbon emissions by performing rolling optimization of energy storage charging and discharging power, photovoltaic output power, and grid interaction power over multiple future time periods.

4. The system according to claim 1, characterized in that, In the multi-device collaborative optimization module, each device interacts with its operating status and power demand information in real time through a unified data interface protocol to achieve collaborative decision-making.

5. The system according to claim 1, characterized in that, The low-carbon economy dual-objective optimization module introduces a carbon trading mechanism and a carbon emission factor, and uses a fuzzy decision-making method to select the optimal solution from the Pareto frontier solution set.

6. The system according to claim 1, characterized in that, The convolutional neural network fault diagnosis model of the adaptive fault diagnosis and fast recovery module can quickly identify the fault type and location, and automatically perform fault isolation and recovery operations.

7. A method for intelligent regulation and coordinated operation of photovoltaic energy storage microgrids, characterized in that, Includes the following steps:

1. Collect data on sunlight, weather, and load, and use a fusion model of spatiotemporal convolutional neural networks and long short-term memory networks to predict photovoltaic power generation and load demand; 2. Establish a mathematical model of energy flow in a microgrid, with the goal of minimizing economic costs and carbon emissions, and use a model predictive control algorithm to perform rolling optimization allocation of energy flow; 3. A distributed collaborative control architecture is used to achieve coordinated operation of devices within the microgrid based on consensus and distributed optimization algorithms; 4. Construct a dual-objective optimization operation model that includes economic costs and carbon costs, solve it using a multi-objective particle swarm optimization algorithm, and select the optimal operation scheme through fuzzy decision-making; 5. Data is collected using intelligent sensors, and fault diagnosis is performed through convolutional neural networks to achieve fault isolation and rapid system recovery.

8. The method according to claim 7, characterized in that, In the prediction step, the model is trained using historical data and updated every 15 minutes with prediction data for the next 15 minutes to 24 hours.

9. The method according to claim 7, characterized in that, In the energy flow regulation step, the energy allocation scheme for the next 4 hours is optimized in 15-minute intervals, and the control command for the current time period is executed.

10. The method according to claim 7, characterized in that, In the fault diagnosis and recovery steps, after the convolutional neural network model identifies the fault, the system completes fault isolation and power restoration operations within 5 minutes.

Images