Optimization control method and system for flexibility transformation and deep peak regulation of thermal power plant

Abstract

The invention provides a thermal power plant flexibility transformation and deep peak regulation optimization control method and system, and relates to the technical field of control engineering, and the method comprises the steps: collecting the omnibearing operation data of a thermal power plant to construct a multi-level thermal power plant dynamic nonlinear model, carrying out the real-time identification of model parameters, generating an identification result, and building a multi-dimensional thermal power plant operation characteristic database; based on the thermal power plant operation characteristic database, designing a transformation scheme, and performing visual simulation and optimization to obtain a flexible transformation result; constructing a multi-objective optimization function, formulating an adaptive deep peak regulation optimization control strategy, carrying out short-term and medium-term prediction on a power grid load, generating an initial control strategy based on a prediction result, carrying out online optimization and adjustment, and continuously evaluating and optimizing the control strategy.

Description

Technical Field

[0001] The present invention relates to the field of control engineering technology, and in particular to a method and system for optimizing control of flexibility transformation and deep peak regulation of a thermal power plant. Background Art

[0002] As the proportion of renewable energy in the power system continues to increase, traditional thermal power plants are facing unprecedented challenges. On the one hand, thermal power plants need to have higher flexibility to cope with rapid fluctuations in grid load. On the other hand, thermal power plants also need to meet increasingly stringent environmental protection requirements while ensuring safe and economical operation.

[0003] Existing thermal power plant control systems and operation strategies have limited peak-shaving capabilities, low operating efficiency and poor environmental performance when dealing with these challenges. At the same time, frequent peak-shaving will aggravate equipment fatigue damage, reduce equipment life and increase maintenance costs.

[0004] Therefore, a solution is urgently needed to solve the problems existing in the prior art. Summary of the invention

[0005] The embodiments of the present invention provide a method and system for controlling flexibility transformation and deep peak regulation of a thermal power plant, which can at least solve some of the problems existing in the prior art.

[0006] A first aspect of an embodiment of the present invention provides a method for flexibility transformation and deep peak load optimization control of a thermal power plant, comprising:

[0007] A distributed sensor network is used to collect all-round operation data of the thermal power plant, and edge computing technology is used to pre-process the collected data, including outlier detection, data smoothing and compression. Based on the pre-processed data, a multi-level dynamic nonlinear model of the thermal power plant is constructed, including the fuel delivery system, boiler system, turbine system, generator system and pollutant control system. A hybrid deep learning algorithm is used, combined with a long short-term memory network and a convolutional neural network, to train and optimize the models of each subsystem. The improved adaptive quantum behavior particle swarm optimization algorithm is used to identify the model parameters in real time, generate identification results, and based on the identification results, a multi-dimensional thermal power plant operation characteristic database is established, including efficiency curves, emission characteristics and dynamic responses of the start-stop process under different loads;

[0008] Based on the operation characteristic database of thermal power plants, data mining technology is used to analyze the operation bottlenecks of the units, including load change rate limiting factors and minimum technical output constraints, and multi-objective and multi-scheme flexibility transformation plans are designed, including the design of intelligent ash cleaning systems for boiler dust prevention, intelligent optimization of turbine bypass systems, frequency conversion transformation of water supply systems, and transformation of combustion optimization systems. Virtual reality and augmented reality technologies are used to visualize and optimize the transformation plans. During the transformation implementation process, digital twin technology based on edge computing is used to build a high-fidelity digital model of the thermal power plant. Through the transformation, the start-stop speed, load change rate and minimum technical output of the units are improved, and the flexibility transformation results are obtained;

[0009] A multi-objective optimization function is constructed based on the dynamic nonlinear model and the results of flexibility transformation. The objectives include maximizing unit efficiency, minimizing emissions, maximizing peak-shaving capacity and maximizing equipment life. An improved model predictive control algorithm is used in combination with a deep reinforcement learning method to formulate an adaptive deep peak-shaving optimization control strategy. Short-term and medium-term forecasts of grid load are made. Based on the prediction results, a deep deterministic policy gradient algorithm is used to generate an initial control strategy. The initial control strategy is optimized and adjusted online. A hierarchical distributed control architecture is designed to generate a global optimization goal. The global goal is decomposed into local goals of each subsystem, and the interactions between subsystems are coordinated. Precise control of each subsystem is achieved through intelligent actuators to achieve coordinated control of the unit. A real-time performance evaluation mechanism based on big data analysis is established. A sliding time window method is used to continuously evaluate and optimize the control strategy. Based on the evaluation results, an adaptive learning algorithm is used to continuously optimize the control strategy.

[0010] In an optional embodiment,

[0011] A distributed sensor network is used to collect all-round operation data of the thermal power plant, and the edge computing technology is used to pre-process the collected data, including outlier detection, data smoothing and compression. Based on the pre-processed data, a multi-level dynamic nonlinear model of the thermal power plant is constructed, including the fuel delivery system, boiler system, turbine system, generator system and pollutant control system. A hybrid deep learning algorithm is used, combined with long short-term memory network and convolutional neural network, to train and optimize the models of each subsystem. The improved adaptive quantum behavior particle swarm optimization algorithm is used to identify the model parameters in real time, generate identification results and establish a multi-dimensional thermal power plant operation characteristic database based on the identification results, including efficiency curves, emission characteristics and start-stop process dynamic responses under different loads.

[0012] High-precision sensors are placed on the steam turbines, boilers and generators of thermal power plants to collect real-time operating parameters, and edge computing nodes are deployed on site. The edge computing nodes are equipped with ARM Cortex-A72 processors and 4GB RAM, running a lightweight Linux operating system;

[0013] Using the edge computing node to preprocess the collected data, the preprocessing includes using an improved local anomaly factor algorithm to perform outlier detection, the improved local anomaly factor algorithm reduces the computational complexity by combining a sliding time window technology, using an adaptive Savitzky-Golay filter to perform data smoothing, the adaptive Savitzky-Golay filter automatically adjusts the filter window size and the polynomial order according to the local characteristics of the data, and using a hybrid compression algorithm based on wavelet transform and deep autoencoder to perform data compression to obtain preprocessed data;

[0014] Based on the preprocessed data, a multi-level dynamic nonlinear model of a thermal power plant is constructed, the model includes five subsystems: a fuel delivery system, a boiler system, a steam turbine system, a generator system and a pollutant control system;

[0015] A hybrid deep learning algorithm is used to train and optimize each subsystem model. The hybrid deep learning algorithm combines a long short-term memory network and a convolutional neural network, wherein the long short-term memory network adopts a bidirectional structure, the convolutional neural network adopts a multi-scale convolution kernel, and the outputs of the long short-term memory network and the convolutional neural network are fused through an attention mechanism;

[0016] An improved adaptive quantum behavior particle swarm optimization algorithm is used to realize real-time identification of model parameters, quantum states are used to represent particle positions and update phase angles, and quantum revolving gates are used to realize particle motion to obtain model identification results. Based on the model identification results, a multi-dimensional thermal power plant operation characteristic database is constructed. The database adopts a distributed storage architecture, uses Apache Cassandra as the underlying storage system, and develops a data access interface based on RESTful API, which supports multi-parameter joint queries and time series slicing. AES-256 is used to encrypt and store data, implement role-based access control, and record audit logs for data access operations.

[0017] In an optional embodiment,

[0018] Using quantum states to represent particle positions and update phase angles is shown in the following formula:

[0019] Δθ=ωΔθ prev +c1r1(θ pbest -θ)+c2r2(θ gbest -θ)+α·(rand()-0.5)·π;

[0020] Among them, △θ is the change of phase angle, which indicates the update amplitude of the current phase angle, ω indicates the inertia weight, and △θ prevrepresents the phase angle change of the previous step, c1 represents the individual learning factor, r1 represents a random number between 0 and 1, θ pbest represents the individual optimal phase angle, c2 represents the global learning factor, r2 represents another random number between 0 and 1, θ gbest represents the global optimal phase angle, α represents the interference intensity, and rand() represents the uncertainty function, which is used to randomly generate a random number between 0 and 1.

[0021] In an optional embodiment,

[0022] Based on the operation characteristic database of thermal power plants, data mining technology is used to analyze the operation bottlenecks of the units, including load change rate limiting factors and minimum technical output constraints, and multi-objective and multi-scheme flexibility transformation plans are designed, including the design of boiler anti-ashing intelligent cleaning system, intelligent optimization of turbine bypass system, frequency conversion transformation of water supply system and combustion optimization system transformation. Virtual reality and augmented reality technologies are used to visualize and optimize the transformation plan. During the transformation implementation process, digital twin technology based on edge computing is used to build a high-fidelity digital model of the thermal power plant. The start-stop speed, load change rate and minimum technical output of the units are improved through transformation. The flexibility transformation results include:

[0023] Constructing a dynamic nonlinear model of the boiler system, and developing an intelligent combustion optimization system based on the dynamic nonlinear model of the boiler system, wherein the intelligent combustion optimization system includes using a long short-term memory network to predict future load changes, using an improved nonlinear model predictive control algorithm to optimize the coal feed rate, primary air volume and secondary air volume in real time, using a multi-objective optimization strategy to construct an optimization objective function, and using a sequential quadratic programming algorithm combined with a trust region method to solve nonlinear optimization problems;

[0024] Construct a steam turbine system model, and develop an adaptive sliding pressure operation control strategy based on the steam turbine system model, wherein the adaptive sliding pressure operation control strategy includes training the control strategy using a deep deterministic policy gradient algorithm, defining a state space including main steam pressure, temperature, flow rate and generator output power, defining an action space as parameters of a sliding pressure curve and designing a reward function, deploying edge computing nodes on feedwater pumps, induced draft fans and forced draft fans, running a lightweight model predictive control algorithm on each edge computing node, designing a distributed coordination mechanism based on federated learning, wherein each edge node maintains a local model, updates model parameters using local data, and sends the updated model parameters to a central server, wherein the central server aggregates all updates and sends the aggregated updates back to each edge node, and the edge node updates the local model using the received updates;

[0025] Develop a comprehensive intelligent dispatching system, treat the boiler system, turbine system and auxiliary system as independent agents, adopt an improved multi-agent soft actor-critic algorithm to achieve overall optimal control, each agent has an agent strategy network and value function network corresponding to the current agent, introduce a global value function to evaluate the value of joint actions, and make decisions based on local value functions and global value functions;

[0026] Visual simulation is carried out to construct a high-fidelity digital model of the thermal power plant. Through transformation, the start-stop speed, load change rate and minimum technical output of the units are improved to obtain flexible transformation results.

[0027] In an optional embodiment,

[0028] The update target of the value function network is shown in the following formula:

[0029]

[0030] Among them, E represents expectation, s t Indicates the current state, a t Indicates the current action, r t represents the current reward value, s t+1 represents the next state, D represents the experience replay buffer, Qi(s t , a t ) represents the value function network for the current state s t and the current action a t The estimated value of , γ represents the discount factor, According to the strategy π i From the next state s t+1 Randomly select action a t+1 The expected value, Q i (s t+1 , a t+1 ) represents the value function network for the next state s t+1和 Next action a t+1 The estimated value of , β represents the policy entropy adjustment factor, and P represents the current policy π i In status t+1 Next select action a t+1 The logarithmic probability of .

[0031] In an optional embodiment,

[0032] According to the dynamic nonlinear model and the results of flexibility transformation, a multi-objective optimization function is constructed. The objectives include maximizing unit efficiency, minimizing emissions, maximizing peak-shaving capacity, and maximizing equipment life. An improved model predictive control algorithm is used, combined with a deep reinforcement learning method, to formulate an adaptive deep peak-shaving optimization control strategy. Short-term and medium-term forecasts of power grid load are made. Based on the forecast results, a deep deterministic policy gradient algorithm is used to generate an initial control strategy. The initial control strategy is optimized and adjusted online. A hierarchical distributed control architecture is designed to generate a global optimization goal. The global goal is decomposed into local goals of each subsystem, and the interactions between subsystems are coordinated. Accurate control of each subsystem is achieved through intelligent actuators to achieve coordinated control of the unit. A real-time performance evaluation mechanism based on big data analysis is established. A sliding time window method is used to continuously evaluate and optimize the control strategy. Based on the evaluation results, an adaptive learning algorithm is used to continuously optimize the control strategy, including:

[0033] Based on the dynamic nonlinear model and the flexibility transformation results, a multi-objective optimization function is constructed, wherein the objectives include maximizing unit efficiency, minimizing emissions, maximizing peak load regulation capacity, and maximizing equipment life, and an improved model predictive control algorithm is designed, wherein the improved model predictive control algorithm includes a state estimator, a constraint processor, and an optimization solver;

[0034] Constructing a deep reinforcement learning framework, the deep reinforcement learning framework includes a deep policy network and a value network, the deep policy network is used to generate control actions, and the value network is used to evaluate state values, and combining the improved model predictive control algorithm with the deep reinforcement learning framework to form an adaptive deep peak-shaving optimization control strategy;

[0035] A method combining a long short-term memory network and a convolutional neural network is used to perform short-term and medium-term forecasts on the power grid load to obtain load forecast results. Based on the load forecast results, an initial control strategy is generated according to the improved model predictive control algorithm and a deep deterministic policy gradient algorithm. The deep deterministic policy gradient algorithm includes an actor network and a critic network. Both the actor network and the critic network adopt a four-layer fully connected neural network structure.

[0036] Design an online optimization module to optimize and adjust the initial control strategy in real time. The online optimization module adopts a model predictive control method to solve the optimal control sequence in the rolling time domain and construct a hierarchical distributed control architecture. The hierarchical distributed control architecture includes a global coordination layer and a local control layer. The global coordination layer is responsible for generating the global optimization target. The decomposition and coordination algorithm is used to decompose the global target into the local targets of each subsystem. The local control layer is responsible for executing the precise control of each subsystem. The adaptive PID control algorithm is used to track the local targets. Design an intelligent actuator. The intelligent actuator includes an advanced actuator and an embedded controller to achieve precise control of each subsystem. Establish a real-time performance evaluation mechanism based on big data analysis. Use a sliding time window method to continuously evaluate and optimize the adaptive deep peak-shaving optimization control strategy.

[0037] Based on the evaluation results of the real-time performance evaluation mechanism, an adaptive learning algorithm is used to continuously optimize the adaptive deep peak-shaving optimization control strategy. The adaptive learning algorithm uses the gradient descent method to update the control strategy parameters. A safety monitoring module is set to monitor key operating parameters in real time. When an anomaly is detected or a constraint is about to be violated, a preset safety protection strategy is triggered.

[0038] A second aspect of an embodiment of the present invention provides a thermal power plant flexibility transformation and deep peak load optimization control system, comprising:

[0039] The first unit is used to collect all-round operation data of thermal power plants using distributed sensor networks, pre-process the collected data using edge computing technology, including outlier detection, data smoothing and compression, and build a multi-level dynamic nonlinear model of thermal power plants based on the pre-processed data, including fuel delivery system, boiler system, turbine system, generator system and pollutant control system. A hybrid deep learning algorithm is used in combination with long short-term memory network and convolutional neural network to train and optimize the models of each subsystem. The improved adaptive quantum behavior particle swarm optimization algorithm is used to identify the model parameters in real time, generate identification results and establish a multi-dimensional thermal power plant operation characteristic database based on the identification results, including efficiency curves, emission characteristics and dynamic response of start-stop process under different loads;

[0040] The second unit is used to analyze the operating bottlenecks of the units based on the operating characteristics database of the thermal power plant using data mining technology, including load change rate limiting factors and minimum technical output constraints, and design multi-objective and multi-scheme flexibility transformation plans, including the design of the intelligent ash cleaning system for boiler anti-ash accumulation, intelligent optimization of the turbine bypass system, frequency conversion transformation of the water supply system and transformation of the combustion optimization system. Virtual reality and augmented reality technologies are used to visualize and optimize the transformation plan. During the transformation implementation process, the digital twin technology based on edge computing is used to build a high-fidelity digital model of the thermal power plant. Through the transformation, the start-stop speed, load change rate and minimum technical output of the units are improved, and the flexibility transformation results are obtained;

[0041] The third unit is used to construct a multi-objective optimization function based on the dynamic nonlinear model and flexibility transformation results. The objectives include maximizing unit efficiency, minimizing emissions, maximizing peak-shaving capacity and maximizing equipment life. An improved model predictive control algorithm is used in combination with a deep reinforcement learning method to formulate an adaptive deep peak-shaving optimization control strategy. Short-term and medium-term forecasts of grid load are made. Based on the forecast results, a deep deterministic policy gradient algorithm is used to generate an initial control strategy. The initial control strategy is optimized and adjusted online. A hierarchical distributed control architecture is designed to generate a global optimization goal. The global goal is decomposed into local goals of each subsystem, and the interactions between subsystems are coordinated. Precise control of each subsystem is achieved through intelligent actuators to achieve coordinated control of the unit. A real-time performance evaluation mechanism based on big data analysis is established. A sliding time window method is used to continuously evaluate and optimize the control strategy. Based on the evaluation results, an adaptive learning algorithm is used to continuously optimize the control strategy.

[0042] According to a third aspect of the embodiments of the present invention,

[0043] An electronic device is provided, comprising:

[0044] processor;

[0045] a memory for storing processor-executable instructions;

[0046] The processor is configured to call the instructions stored in the memory to execute the aforementioned method.

[0047] According to a fourth aspect of the embodiments of the present invention,

[0048] A computer-readable storage medium is provided, on which computer program instructions are stored. When the computer program instructions are executed by a processor, the aforementioned method is implemented.

[0049] In the present invention, through multi-objective optimization and deep reinforcement learning, the load change rate of the unit is significantly improved, the minimum technical output is reduced, the start-up and shutdown speeds are enhanced, the peak-shaving capability and system flexibility of the thermal power unit are greatly improved, and the improved model predictive control algorithm and deep learning technology are adopted to achieve the optimized operation of the unit under different loads, significantly improve the unit efficiency, reduce the power generation cost and fuel consumption, and effectively reduce the loss of equipment and extend the expected life of major equipment by incorporating the maximization of equipment life into the optimization target and adopting an evaluation method based on a fatigue damage model. The response speed of the system is significantly improved through the cooperation of online optimization modules and intelligent actuators, which is crucial for rapid peak-shaving and frequency regulation. In summary, the present invention not only optimizes the performance of a single unit, but also provides strong support for the flexibility, reliability and cleanliness of the entire power system, which is of great significance to promoting energy transformation and sustainable development. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 It is a flow chart of a method for optimizing control of flexibility transformation and deep peak load regulation of a thermal power plant according to an embodiment of the present invention;

[0051] Figure 2 It is a structural schematic diagram of a thermal power plant flexibility transformation and deep peak-shaving optimization control system according to an embodiment of the present invention. DETAILED DESCRIPTION

[0052] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0053] The technical solution of the present invention is described in detail with specific embodiments below. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described in detail in some embodiments.

[0054] Figure 1 Schematic diagram of the process of the embodiment of the present invention, such as Figure 1 As shown, the method includes:

[0055] S1. Use a distributed sensor network to collect all-round operation data of the thermal power plant, use edge computing technology to pre-process the collected data, including outlier detection, data smoothing and compression, and build a multi-level dynamic nonlinear model of the thermal power plant based on the pre-processed data, including the fuel delivery system, boiler system, turbine system, generator system and pollutant control system. Use a hybrid deep learning algorithm, combined with long short-term memory network and convolutional neural network, to train and optimize the models of each subsystem, use an improved adaptive quantum behavior particle swarm optimization algorithm to identify the model parameters in real time, generate identification results and establish a multi-dimensional thermal power plant operation characteristics database based on the identification results, including efficiency curves, emission characteristics and dynamic response of the start-stop process under different loads;

[0056] The hybrid deep learning algorithm combines multiple deep learning techniques (such as convolutional neural networks, recursive neural networks, etc.) to improve the performance and generalization ability of the model and can more effectively capture complex data features. The improved adaptive quantum behavior particle swarm optimization algorithm is a variant of particle swarm optimization, which combines the idea of ​​quantum computing. By introducing quantum bits and quantum rotating gates, the search behavior of particles is more flexible, thereby improving the global search ability and convergence speed. The multi-dimensional thermal power plant operation characteristics database is a data management system that integrates multiple operating parameters and performance indicators, and is designed to support the optimization and decision-making of thermal power plants.

[0057] In an optional embodiment,

[0058] A distributed sensor network is used to collect all-round operation data of the thermal power plant, and the edge computing technology is used to pre-process the collected data, including outlier detection, data smoothing and compression. Based on the pre-processed data, a multi-level dynamic nonlinear model of the thermal power plant is constructed, including the fuel delivery system, boiler system, turbine system, generator system and pollutant control system. A hybrid deep learning algorithm is used, combined with long short-term memory network and convolutional neural network, to train and optimize the models of each subsystem. The improved adaptive quantum behavior particle swarm optimization algorithm is used to identify the model parameters in real time, generate identification results and establish a multi-dimensional thermal power plant operation characteristic database based on the identification results, including efficiency curves, emission characteristics and start-stop process dynamic responses under different loads.

[0059] High-precision sensors are placed on the steam turbines, boilers and generators of thermal power plants to collect real-time operating parameters, and edge computing nodes are deployed on site. The edge computing nodes are equipped with ARM Cortex-A72 processors and 4GB RAM, running a lightweight Linux operating system;

[0060] Using the edge computing node to preprocess the collected data, the preprocessing includes using an improved local anomaly factor algorithm to perform outlier detection, the improved local anomaly factor algorithm reduces the computational complexity by combining a sliding time window technology, using an adaptive Savitzky-Golay filter to perform data smoothing, the adaptive Savitzky-Golay filter automatically adjusts the filter window size and the polynomial order according to the local characteristics of the data, and using a hybrid compression algorithm based on wavelet transform and deep autoencoder to perform data compression to obtain preprocessed data;

[0061] Based on the preprocessed data, a multi-level dynamic nonlinear model of a thermal power plant is constructed, the model includes five subsystems: a fuel delivery system, a boiler system, a steam turbine system, a generator system and a pollutant control system;

[0062] A hybrid deep learning algorithm is used to train and optimize each subsystem model. The hybrid deep learning algorithm combines a long short-term memory network and a convolutional neural network, wherein the long short-term memory network adopts a bidirectional structure, the convolutional neural network adopts a multi-scale convolution kernel, and the outputs of the long short-term memory network and the convolutional neural network are fused through an attention mechanism;

[0063] An improved adaptive quantum behavior particle swarm optimization algorithm is used to realize real-time identification of model parameters, quantum states are used to represent particle positions and update phase angles, and quantum revolving gates are used to realize particle motion to obtain model identification results. Based on the model identification results, a multi-dimensional thermal power plant operation characteristic database is constructed. The database adopts a distributed storage architecture, uses Apache Cassandra as the underlying storage system, and develops a data access interface based on RESTful API, which supports multi-parameter joint queries and time series slicing. AES-256 is used to encrypt and store data, implement role-based access control, and record audit logs for data access operations.

[0064] The adaptive Savitzky-Golay filter is a signal processing technology suitable for denoising and data smoothing. The hybrid compression algorithm based on wavelet transform and deep autoencoder combines the time-frequency analysis capability of wavelet transform with the feature extraction capability of deep autoencoder, aiming to improve data compression efficiency, especially when processing complex signals (such as images or audio). The quantum state refers to the state of the quantum system, which is described by wave function or density matrix, reflecting the physical properties of the system. The phase angle refers to the quantum state in quantum computing, which can be expressed as a complex form, where the phase angle is an important parameter, which determines the position of the quantum state on the complex plane and affects the result of the quantum operation. The quantum rotation gate is one of the basic operations in quantum computing, and the operation of quantum bits is realized by rotating the phase angle of the quantum state. Common ones are RX, RY and RZ gates, corresponding to rotations in different axes respectively.

[0065] High-precision sensors are placed on key equipment in thermal power plants. Vibration sensors, temperature sensors and pressure sensors are installed on steam turbines to monitor parameters such as bearing vibration, steam temperature and pressure. Oxygen content sensors, temperature sensors and flow sensors are installed on boilers to monitor parameters such as combustion efficiency, superheated steam temperature and feed water flow. Voltage sensors, current sensors and power factor sensors are installed on generators to monitor parameters such as output voltage, current and power factor. The sampling frequency of these sensors is set to 100 Hz to ensure that rapid changes in the operating status of the equipment are captured.

[0066] Next, edge computing nodes were deployed on site, with one edge computing node deployed in each major equipment area (such as the turbine room, boiler area, and generator room). Each node was equipped with an ARM Cortex-A72 processor with a main frequency of 2.5GHz, 4GB DDR4 RAM, and 128GB SSD storage. The node ran a customized lightweight Linux operating system based on Ubuntu 20.04LTS, retaining only the necessary system components and drivers to reduce system overhead and improve operating efficiency.

[0067] The edge computing node preprocesses the collected data and uses the improved local anomaly factor algorithm to detect outliers. The sliding time window technology is adopted. The window size is set to 1 minute and slides every 10 seconds. For each data point, its local density ratio with other data points in the window is calculated. If the ratio exceeds the preset threshold (such as 3.0), it is marked as an outlier.

[0068] Adaptive Savitzky-Golay filter is used for data smoothing. The initial window size of the filter is set to 21 data points and the polynomial order is 3. The filter dynamically adjusts the window size by calculating the local variance of the data in the window. If the local variance increases, the window size is reduced to retain details; if the local variance decreases, the window size is increased to enhance the smoothing effect. The polynomial order is also adjusted according to the complexity of the data, ranging from 2 to 5.

[0069] A hybrid compression algorithm based on wavelet transform and deep autoencoder is used for data compression. First, the data is subjected to discrete wavelet transform, and the Daubechies 4 wavelet basis is used with a decomposition layer number of 3. The wavelet coefficients are input into the deep autoencoder. The encoder part of the autoencoder contains 3 hidden layers with 256, 128 and 64 neurons respectively. The decoder part is symmetrical and uses the ReLU activation function and batch normalization layer to improve training efficiency and generalization ability. In this way, the original data can be compressed to 1 / 10 of its original size while keeping the reconstruction error below 1%;

[0070] Based on the preprocessed data, a multi-level dynamic nonlinear model of a thermal power plant is constructed. The model includes five subsystems: fuel delivery system, boiler system, steam turbine system, generator system and pollutant control system. Each subsystem model considers the nonlinear relationship and dynamic characteristics between key parameters. For example, the boiler system model includes the combustion process, water wall heat transfer, superheater and reheater, and considers the influence of factors such as fuel composition, feed water flow, and steam parameters.

[0071] A hybrid deep learning algorithm is used to train and optimize each subsystem model. The algorithm combines the long short-term memory network and the convolutional neural network. The long short-term memory network adopts a bidirectional structure with 128 hidden layer neurons to capture the long-term dependencies of time series data. The convolutional neural network uses multi-scale convolution kernels, including convolution kernels of three sizes: 1x1, 3x3, and 5x5. Each size has 32 filters to extract multi-scale features. The outputs of the long short-term memory network and the convolutional neural network are fused through the attention mechanism. The attention weight is calculated by the softmax function. The Adam optimizer is used for model training, the learning rate is set to 0.001, the batch size is 64, and the number of training rounds is 100.

[0072] An improved adaptive quantum behavior particle swarm optimization algorithm is used to realize real-time identification of model parameters. The particle swarm size is set to 50, the maximum number of iterations is 100, each particle uses a quantum bit to represent its position, the initial phase angle is randomly generated, and the particle motion is achieved through a quantum rotating gate. The rotation angle is calculated based on the current position of the particle, the individual optimal position, and the global optimal position. An adaptive inertia weight is introduced with an initial value of 0.9, which decreases linearly to 0.4 as the number of iterations increases to balance the global search and local search capabilities.

[0073] Based on the model identification results, a multi-dimensional thermal power plant operation characteristics database is constructed. The database adopts a distributed storage architecture and uses Apache Cassandra as the underlying storage system. Three data nodes are set up, each node is equipped with an 8-core CPU, 32GB RAM and 1TB SSD storage. The data is partitioned according to timestamp and device ID to optimize query performance.

[0074] Develop a data access interface based on RESTful API, supporting multi-parameter joint query and time series slicing. The API is implemented using Python Flask framework and deployed on a separate application server;

[0075] The stored data is encrypted using the AES-256 algorithm. The key length is 256 bits, encrypted using CBC mode, and padded using PKCS7. Key management is protected by a hardware security module (HSM) to implement role-based access control, defining three roles: administrator, engineer, and operator, each with different data access permissions. For example, operators can only view real-time operating data, while engineers can access historical data and perform advanced analysis.

[0076] In this embodiment, by arranging high-precision sensors on key equipment, the core operating parameters of the thermal power plant are accurately captured, providing high-quality raw data for subsequent data analysis and model building. By performing preliminary data preprocessing on site, the amount of data transmission is greatly reduced, the network bandwidth requirement is reduced, and the computing burden of the central server is also reduced. The distributed computing architecture enables the system to respond to changes in equipment status more quickly, providing possibilities for real-time control and fault warning. The hybrid compression algorithm based on wavelet transform and deep autoencoder significantly reduces storage and transmission costs while maintaining a low reconstruction error. The dynamic nonlinear model can better capture the transient characteristics and nonlinear behavior of the system, providing a more reliable theoretical basis for optimizing control and fault diagnosis. In summary, this embodiment realizes the efficient collection, processing, storage and analysis of thermal power plant operation data, provides strong technical support for the intelligent and refined management of thermal power plants, and helps to improve the overall operating efficiency and economic benefits of thermal power plants.

[0077] In an optional embodiment,

[0078] Using quantum states to represent particle positions and update phase angles is shown in the following formula:

[0079] Δθ=ωΔθ prev +c1r1(θ pbest -θ)+c2r2(θ gbest -θ)+α·(rand()-0.5)·π;

[0080] Among them, △θ is the change of phase angle, which indicates the update amplitude of the current phase angle, ω indicates the inertia weight, and △θ prev represents the phase angle change of the previous step, c1 represents the individual learning factor, r1 represents a random number between 0 and 1, θ pbest represents the individual optimal phase angle, c2 represents the global learning factor, r2 represents another random number between 0 and 1, θ gbest represents the global optimal phase angle, α represents the interference intensity, and rand() represents the uncertainty function, which is used to randomly generate a random number between 0 and 1.

[0081] In this embodiment, quantum states are used to represent particle positions, allowing particles to explore multiple possible solutions at the same time, so that the algorithm can conduct a wider and deeper search in the solution space, greatly improving the probability of finding the global optimal solution. The superposition property of quantum states enables particles to evaluate multiple potential solutions at the same time in each iteration, thereby speeding up the process of finding the optimal solution and achieving better optimization results within a smaller number of iterations. The inertia term maintains the movement trend of the particles, which is helpful for global search. Individual cognitive terms and social cognitive terms guide particles to move toward known good solutions and promote local search. Random perturbation terms increase the diversity of search and help to jump out of local optimality. The adaptive mechanism enables the algorithm to maintain good performance at different stages of the optimization process, and can conduct extensive global searches in the early stages, and refine local searches in the later stages. In summary, this embodiment provides a more efficient and accurate solution for parameter identification of dynamic nonlinear models of thermal power plants, thereby laying a solid foundation for intelligent operation and optimization control of thermal power plants.

[0082] S2. Based on the operation characteristic database of thermal power plants, data mining technology is used to analyze the operation bottlenecks of the units, including load change rate limiting factors and minimum technical output constraints, and multi-objective and multi-scheme flexibility transformation plans are designed, including the design of intelligent ash cleaning system for boiler anti-ash accumulation, intelligent optimization of turbine bypass system, frequency conversion transformation of water supply system and transformation of combustion optimization system. Virtual reality and augmented reality technologies are used to visualize and optimize the transformation plan. During the transformation implementation process, digital twin technology based on edge computing is used to build a high-fidelity digital model of the thermal power plant. Through the transformation, the start-stop speed, load change rate and minimum technical output of the units are improved, and the flexibility transformation results are obtained;

[0083] The data mining is the process of extracting valuable information from large amounts of data. Common techniques include classification, clustering, association rule mining and regression analysis. The minimum technical output constraint refers to setting the minimum output level that the power generation equipment must reach when performing power generation scheduling or resource allocation to ensure the stability and reliability of the system. The edge computing-based digital twin technology refers to using edge computing capabilities to monitor and optimize virtual models of physical systems in real time. Digital twins create digital copies of physical devices or systems, collect and analyze data in real time, and thus reflect their status and behavior. Edge computing processes data at the location where the data is generated (such as next to the device) to reduce latency and increase response speed.

[0084] In an optional embodiment,

[0085] Based on the operation characteristic database of thermal power plants, data mining technology is used to analyze the operation bottlenecks of the units, including load change rate limiting factors and minimum technical output constraints, and multi-objective and multi-scheme flexibility transformation plans are designed, including the design of boiler anti-ashing intelligent cleaning system, intelligent optimization of turbine bypass system, frequency conversion transformation of water supply system and combustion optimization system transformation. Virtual reality and augmented reality technologies are used to visualize and optimize the transformation plan. During the transformation implementation process, digital twin technology based on edge computing is used to build a high-fidelity digital model of the thermal power plant. The start-stop speed, load change rate and minimum technical output of the units are improved through transformation. The flexibility transformation results include:

[0086] Constructing a dynamic nonlinear model of the boiler system, and developing an intelligent combustion optimization system based on the dynamic nonlinear model of the boiler system, wherein the intelligent combustion optimization system includes using a long short-term memory network to predict future load changes, using an improved nonlinear model predictive control algorithm to optimize the coal feed rate, primary air volume and secondary air volume in real time, using a multi-objective optimization strategy to construct an optimization objective function, and using a sequential quadratic programming algorithm combined with a trust region method to solve nonlinear optimization problems;

[0087] Construct a steam turbine system model, and develop an adaptive sliding pressure operation control strategy based on the steam turbine system model, wherein the adaptive sliding pressure operation control strategy includes training the control strategy using a deep deterministic policy gradient algorithm, defining a state space including main steam pressure, temperature, flow rate and generator output power, defining an action space as parameters of a sliding pressure curve and designing a reward function, deploying edge computing nodes on feedwater pumps, induced draft fans and forced draft fans, running a lightweight model predictive control algorithm on each edge computing node, designing a distributed coordination mechanism based on federated learning, wherein each edge node maintains a local model, updates model parameters using local data, and sends the updated model parameters to a central server, wherein the central server aggregates all updates and sends the aggregated updates back to each edge node, and the edge node updates the local model using the received updates;

[0088] Develop a comprehensive intelligent dispatching system, treat the boiler system, turbine system and auxiliary system as independent agents, adopt an improved multi-agent soft actor-critic algorithm to achieve overall optimal control, each agent has an agent strategy network and value function network corresponding to the current agent, introduce a global value function to evaluate the value of joint actions, and make decisions based on local value functions and global value functions;

[0089] Visual simulation is carried out to construct a high-fidelity digital model of the thermal power plant. Through transformation, the start-stop speed, load change rate and minimum technical output of the units are improved to obtain flexible transformation results.

[0090] The dynamic nonlinear model of the boiler system describes the dynamic behavior of the boiler under different operating conditions, including the combustion process, heat exchange and pressure changes, etc. The intelligent combustion optimization system refers to the use of data-driven methods and intelligent algorithms to optimize the combustion process to achieve the goals of improving combustion efficiency and reducing emissions. The coal feed refers to the supply of fuel (such as coal) in the boiler system, which is an important parameter affecting combustion efficiency and boiler operation. The sequential quadratic programming algorithm is an optimization algorithm that is often used to solve optimization problems with quadratic objective functions and linear constraints. The adaptive sliding pressure operation control strategy is a control strategy that optimizes operating efficiency and stability by adjusting the system's operating points (such as pressure and flow). The improved multi-agent soft actor-commentator algorithm is an algorithm in reinforcement learning that optimizes the decision-making process through collaboration and competition among multiple agents.

[0091] A dynamic nonlinear model of the boiler system is constructed, taking into account the complex thermodynamic processes and chemical reactions inside the boiler, including the combustion process, water circulation system, steam system, etc. The model uses a partitioning method to divide the boiler into several control volumes, and establishes mass balance, energy balance and momentum balance equations in each control volume. Considering the nonlinear characteristics of the boiler system, a neural network is introduced to describe the parts that are difficult to express with a mechanism model, such as the complex relationship between combustion efficiency and multiple factors;

[0092] Based on the constructed dynamic nonlinear model of the boiler system, an intelligent combustion optimization system is developed, and a long short-term memory network is used to predict future load changes. The input of the long short-term memory network includes historical load data, weather forecast information, holiday information, etc., and the output is the load forecast for the next 24 hours. The network structure includes a long short-term memory network layer with 128 neurons, a fully connected layer with 64 neurons and an output layer. The Adam optimizer is used to train the network, the learning rate is set to 0.001, and the batch size is 32;

[0093] Based on the load forecast results, the improved nonlinear model predictive control algorithm is used to optimize the coal feed rate, primary air volume and secondary air volume in real time. The improved nonlinear model predictive control algorithm takes into account the strong coupling nonlinear characteristics of the boiler system. The prediction time domain is set to 30 minutes and the control time domain is 5 minutes. The constraints of the control variables include the coal feed rate between 30-100t / h and the primary air volume between 100-300kNm 3 / h, secondary air volume is 50-150kNm 3 / h, the optimization objective function adopts a multi-objective optimization strategy, comprehensively considering boiler efficiency, NOx emissions and steam parameter stability;

[0094] The optimization objective function is defined as the weighted sum of boiler efficiency, NOx emissions, and main steam temperature deviation. The weight coefficients are determined by expert experience and historical data analysis, and are 0.5, 0.3, and 0.2, respectively. The sequential quadratic programming algorithm is combined with the trust region method to solve the nonlinear optimization problem. The initial trust region radius of the sequential quadratic programming algorithm is set to 1, the maximum number of iterations is 100, and the convergence tolerance is 1e-6.

[0095] Construct a steam turbine system model. The model includes main components such as high-pressure cylinder, medium-pressure cylinder, low-pressure cylinder, reheater, condenser, etc. Thermodynamic equations and energy balance equations are established for each component, and nonlinear characteristics of the steam turbine are considered, such as valve characteristic curves, efficiency curves, etc. The model input includes main steam parameters, extraction volume, cooling water temperature, etc., and the output is performance indicators such as generator output power and heat rate.

[0096] Based on the steam turbine system model, an adaptive sliding pressure operation control strategy is developed. First, the control strategy is trained using a deep deterministic policy gradient algorithm. The state space is defined to include main steam pressure, temperature, flow, and generator output power. The sampling frequency is 1 Hz. The action space is defined as the parameters of the sliding pressure curve, including the coordinates of the control points of the cubic spline curve. The reward function is designed to consider power generation efficiency, equipment life, and operational stability. Specifically, it is the weighted sum of the power generation efficiency improvement value and the equipment stress penalty term. The weight coefficients are determined to be 0.7 and 0.3 through experiments.

[0097] The deep deterministic policy gradient algorithm uses two neural networks: an actor network and a critic network. The actor network consists of three fully connected layers with 256, 128, and 64 neurons, using the ReLU activation function. The critic network has a similar structure, but merges the action input after the second layer. The Adam optimizer is used with learning rates of 1e-4 and 1e-3, respectively. An experience replay buffer of size 100,000 is used, the batch size is 64, and the target network soft update parameter τ is set to 0.001.

[0098] Edge computing nodes are deployed on the feedwater pump, induced draft fan, and forced draft fan. Each node is equipped with an ARM Cortex-A72 processor and 4GB RAM. A lightweight model predictive control algorithm is run on each edge computing node. The prediction time domain is 10 minutes and the control time domain is 1 minute. The control variables include equipment speed and valve opening. The constraints take into account equipment operation restrictions and process parameter requirements.

[0099] A distributed coordination mechanism based on federated learning is designed. Each edge node maintains a local model and uses local data to update model parameters. The local model uses a three-layer fully connected neural network with 64 and 32 hidden layer neurons and a ReLU activation function. Every 30 minutes, the edge node sends the updated model parameters to the central server. The central server aggregates all updates using a federated averaging algorithm. The aggregation weight is proportional to the amount of data on each node. The aggregated updates are sent back to each edge node, and the edge node uses the received updates to update the local model.

[0100] An integrated intelligent dispatching system is developed, in which the boiler system, turbine system and auxiliary system are regarded as independent agents, and the improved multi-agent soft actor-critic (MA-SAC) algorithm is used to achieve overall optimal control. Each agent corresponds to an agent strategy network and a value function network. The agent strategy network adopts a Gaussian strategy and consists of a three-layer fully connected network with 128 and 64 neurons in the hidden layer. The value function network has a similar structure, but the output layer has only one neuron.

[0101] A global value function is introduced to evaluate the value of joint actions. The global value function network consists of a four-layer fully connected network with 256, 128 and 64 neurons in the hidden layer. Decisions are made based on the local value function and the global value function. A soft Q learning update strategy is used. The initial value of the temperature parameter is set to 0.2 and dynamically adjusted through an automatic adjustment mechanism. Importance sampling technology is used to improve sample efficiency, and the truncation parameter is set to 0.2.

[0102] We conducted visual simulation and constructed a high-fidelity digital model of the thermal power plant. We used the Unity 3D engine to create a three-dimensional scene with a modeling accuracy of centimeters. Through the transformation, we improved the start-up and shutdown speeds, load change rate and minimum technical output of the units, and obtained flexibility transformation results. Specifically, by optimizing the boiler water-cooled wall structure, improving the turbine regulation system and upgrading the DCS control strategy, the unit start-up time was shortened from 4 hours to 2.5 hours, the load change rate was increased from 2% of the rated load / minute to 5% of the rated load / minute, and the minimum technical output was reduced from 40% of the rated load to 25% of the rated load.

[0103] In this embodiment, by constructing a dynamic nonlinear model including a boiler system and a steam turbine system, and combining neural networks to process complex relationships that are difficult to express with a mechanism model, the accuracy of system modeling is greatly improved. Through an adaptive sliding pressure operation control strategy, a deep deterministic policy gradient (DDPG) algorithm is used to train the control strategy, which significantly improves the load regulation capability of the unit. Edge computing nodes are deployed on key equipment to run a lightweight model predictive control algorithm, and global optimization is achieved through a distributed coordination mechanism based on federated learning, which significantly improves the response speed and reliability of the control system. A high-fidelity digital model is constructed with a simulation accuracy of centimeters, which provides a highly realistic virtual environment for equipment transformation and operation optimization, greatly reduces the risk and cost of actual transformation, and speeds up the verification and implementation of optimization schemes. Through a series of transformation measures, the start-stop speed, load change rate and minimum technical output of the unit are significantly improved, so that the thermal power unit can better adapt to the grid fluctuations caused by the large-scale access of renewable energy, and improves the economy and environmental protection of the unit. In summary, this embodiment not only improves the operating efficiency and economic benefits of thermal power plants, but also enhances the adaptability of thermal power units under the background of large-scale access of new energy, and provides strong support for the intelligent and green transformation of thermal power plants.

[0104] In an optional embodiment,

[0105] The update target of the value function network is shown in the following formula:

[0106]

[0107] Among them, E represents expectation, s t Indicates the current state, a t Indicates the current action, r t represents the current reward value, s t+1 represents the next state, D represents the experience replay buffer, Qi(s t , a t ) represents the value function network for the current state s t and the current action a t The estimated value of , γ represents the discount factor, According to the strategy π i From the next state s t+1 Randomly select action a t+1 The expected value, Q i (s t+1 , a t+1 ) represents the value function network for the next state s t+1和 Next action a t+1 The estimated value of , β represents the policy entropy adjustment factor, and P represents the current policy π i In status t+1Next select action a t+1 The logarithmic probability of .

[0108] In this embodiment, the strong correlation between continuous states is broken by introducing the experience replay buffer, the correlation between samples is reduced, the variance in the learning process is effectively reduced, and the stability of the algorithm is improved. The introduction of the policy entropy adjustment factor and the logarithmic probability of the policy helps to avoid the policy from falling into the local optimum and increases the possibility of finding the global optimal solution. The loss is calculated by using the difference between the target network and the main network. This method implements a soft update mechanism that can smooth the update process of the value function, reduce the oscillation in the learning process, and improve the stability and convergence of the algorithm. The dynamic balance mechanism enables the algorithm to adopt the most suitable strategy at different learning stages, thereby improving the learning efficiency. In summary, this embodiment enables it to cope with complex and changeable operating environments more intelligently, efficiently, and flexibly, and provides strong support for the intelligent transformation and green development of thermal power plants.

[0109] S3. Construct a multi-objective optimization function based on the dynamic nonlinear model and flexibility transformation results. The objectives include maximizing unit efficiency, minimizing emissions, maximizing peak-shaving capacity, and maximizing equipment life. Use an improved model predictive control algorithm combined with a deep reinforcement learning method to develop an adaptive deep peak-shaving optimization control strategy. Perform short-term and medium-term forecasts on the grid load. Based on the forecast results, use a deep deterministic policy gradient algorithm to generate an initial control strategy. Perform online optimization and adjustment on the initial control strategy. Design a hierarchical distributed control architecture to generate a global optimization goal. Decompose the global goal into local goals of each subsystem, and coordinate the interactions between subsystems. Use intelligent actuators to achieve precise control of each subsystem to achieve coordinated control of the unit. Establish a real-time performance evaluation mechanism based on big data analysis. Use a sliding time window method to continuously evaluate and optimize the control strategy. Based on the evaluation results, use an adaptive learning algorithm to continuously optimize the control strategy.

[0110] The model predictive control algorithm is a model-based control strategy that optimizes future control inputs by predicting the mathematical model of the system's dynamics. The adaptive deep peak-shaving optimization control strategy combines adaptive control and deep learning methods to optimize the system's load scheduling and peak management. The hierarchical distributed control architecture divides the control system into multiple levels and distributed units, typically including a field level, a control level, and a management level. Each level is responsible for different functions, and the levels coordinate through communication. The real-time performance evaluation mechanism based on big data analysis uses big data analysis technology to monitor and analyze real-time data to evaluate the performance of the system.

[0111] In an optional embodiment,

[0112] According to the dynamic nonlinear model and the results of flexibility transformation, a multi-objective optimization function is constructed. The objectives include maximizing unit efficiency, minimizing emissions, maximizing peak-shaving capacity, and maximizing equipment life. An improved model predictive control algorithm is used, combined with a deep reinforcement learning method, to formulate an adaptive deep peak-shaving optimization control strategy. Short-term and medium-term forecasts of power grid load are made. Based on the forecast results, a deep deterministic policy gradient algorithm is used to generate an initial control strategy. The initial control strategy is optimized and adjusted online. A hierarchical distributed control architecture is designed to generate a global optimization goal. The global goal is decomposed into local goals of each subsystem, and the interactions between subsystems are coordinated. Accurate control of each subsystem is achieved through intelligent actuators to achieve coordinated control of the unit. A real-time performance evaluation mechanism based on big data analysis is established. A sliding time window method is used to continuously evaluate and optimize the control strategy. Based on the evaluation results, an adaptive learning algorithm is used to continuously optimize the control strategy, including:

[0113] Based on the dynamic nonlinear model and the flexibility transformation results, a multi-objective optimization function is constructed, wherein the objectives include maximizing unit efficiency, minimizing emissions, maximizing peak load regulation capacity, and maximizing equipment life, and an improved model predictive control algorithm is designed, wherein the improved model predictive control algorithm includes a state estimator, a constraint processor, and an optimization solver;

[0114] Constructing a deep reinforcement learning framework, the deep reinforcement learning framework includes a deep policy network and a value network, the deep policy network is used to generate control actions, and the value network is used to evaluate state values, and combining the improved model predictive control algorithm with the deep reinforcement learning framework to form an adaptive deep peak-shaving optimization control strategy;

[0115] A method combining a long short-term memory network and a convolutional neural network is used to perform short-term and medium-term forecasts on the power grid load to obtain load forecast results. Based on the load forecast results, an initial control strategy is generated according to the improved model predictive control algorithm and a deep deterministic policy gradient algorithm. The deep deterministic policy gradient algorithm includes an actor network and a critic network. Both the actor network and the critic network adopt a four-layer fully connected neural network structure.

[0116] Design an online optimization module to optimize and adjust the initial control strategy in real time. The online optimization module adopts a model predictive control method to solve the optimal control sequence in the rolling time domain and construct a hierarchical distributed control architecture. The hierarchical distributed control architecture includes a global coordination layer and a local control layer. The global coordination layer is responsible for generating the global optimization target. The decomposition and coordination algorithm is used to decompose the global target into the local targets of each subsystem. The local control layer is responsible for executing the precise control of each subsystem. The adaptive PID control algorithm is used to track the local targets. Design an intelligent actuator. The intelligent actuator includes an advanced actuator and an embedded controller to achieve precise control of each subsystem. Establish a real-time performance evaluation mechanism based on big data analysis. Use a sliding time window method to continuously evaluate and optimize the adaptive deep peak-shaving optimization control strategy.

[0117] Based on the evaluation results of the real-time performance evaluation mechanism, an adaptive learning algorithm is used to continuously optimize the adaptive deep peak-shaving optimization control strategy. The adaptive learning algorithm uses the gradient descent method to update the control strategy parameters. A safety monitoring module is set to monitor key operating parameters in real time. When an anomaly is detected or a constraint is about to be violated, a preset safety protection strategy is triggered.

[0118] The rolling time domain refers to a dynamically updated time range in a control or prediction process, and the sliding time window method is used for data analysis and signal processing by collecting and processing data within a time window of fixed length and continuously moving the window over time.

[0119] Based on the dynamic nonlinear model and the flexibility transformation results, a multi-objective optimization function is constructed. The optimization function comprehensively considers four objectives: maximizing unit efficiency, minimizing emissions, maximizing peak-shaving capacity, and maximizing equipment life. Unit efficiency is measured by heat rate, emissions mainly consider NOx, SO2, and dust, peak-shaving capacity includes load change rate and minimum technical output, and equipment life is evaluated based on fatigue damage models of key components;

[0120] To achieve multi-objective optimization, an improved model predictive control algorithm is designed, which includes three key components: state estimator, constraint processor and optimization solver. The state estimator adopts the extended Kalman filter method to integrate the real-time data from various sensors and accurately estimate the current state of the system. The constraint processor considers various hard and soft constraints of equipment operation, such as boiler steam parameter restrictions and turbine vibration limits. The optimization solver adopts the sequential quadratic programming method to solve the optimal control sequence under the condition of satisfying the constraints.

[0121] A deep reinforcement learning framework is constructed, including a deep policy network and a value network. The deep policy network adopts a four-layer fully connected neural network structure. The number of nodes in the input layer is the system state dimension, the hidden layer has 128 and 64 nodes respectively, and the number of nodes in the output layer is the control action dimension. The value network also adopts a similar structure, but the output layer has only one node, which is used to estimate the state value. Both networks use the ReLU activation function and are trained using the Adam optimizer.

[0122] The improved model predictive control algorithm is combined with the deep reinforcement learning framework to form an adaptive deep peak-shaving optimization control strategy. Specifically, the model predictive control algorithm is used to generate the initial control sequence, which is then used as the starting point for deep reinforcement learning exploration. Deep reinforcement learning continuously optimizes the control strategy and gradually improves the peak-shaving performance by interacting with the environment.

[0123] The method of combining long short-term memory network and convolutional neural network is used to make short-term and medium-term forecasts of power grid load. The long short-term memory network contains two layers of long short-term memory network, each with 128 hidden units, which are used to capture the long-term dependencies of load sequences. The convolutional neural network contains two convolutional layers, using 32 and 64 3x3 convolution kernels respectively, to extract local features of load data. The outputs of the long short-term memory network and the convolutional neural network are connected, and the final load forecasting results are obtained through the fully connected layer.

[0124] Based on the load forecasting results, the initial control strategy is generated using the improved model predictive control algorithm and the deep deterministic policy gradient algorithm. The actor network and the critic network in the deep deterministic policy gradient algorithm both use a four-layer fully connected neural network structure. The input layer of the actor network corresponds to the system state, and the output layer corresponds to the continuous control action. The input of the critic network includes the state and action, and the output is the value estimate of the state-action pair. Both networks use batch normalization technology to accelerate the training process.

[0125] Design an online optimization module to optimize and adjust the initial control strategy in real time. This module uses the model predictive control method to solve the optimal control sequence within a 10-minute rolling time domain and updates the control input every 30 seconds to ensure that the control strategy can respond to system state changes in a timely manner.

[0126] Construct a hierarchical distributed control architecture, including a global coordination layer and a local control layer. The global coordination layer is responsible for generating the global optimization target. The Lagrangian relaxation method is used to decompose the global target into the local targets of each subsystem. The local control layer is responsible for executing the precise control of each subsystem. The adaptive PID control algorithm is used to track the local targets. The PID parameters are adjusted in real time through the fuzzy inference system to adapt to different working conditions.

[0127] Design of intelligent actuators, including advanced actuators and embedded controllers. The actuators use high-precision servo motors with a response time of less than 50 milliseconds. The embedded controller is based on an ARM Cortex-M4 processor running at 168MHz and capable of implementing complex control algorithms.

[0128] Establish a real-time performance evaluation mechanism based on big data analysis, using a 5-minute sliding time window to continuously evaluate the system's efficiency, emissions, peak-shaving performance and other indicators, use principal component analysis to reduce the dimensionality of multidimensional performance indicators, and calculate the comprehensive performance score;

[0129] Based on the evaluation results of the real-time performance evaluation mechanism, an adaptive learning algorithm is used to continuously optimize the adaptive deep peak-shaving optimization control strategy. The adaptive learning algorithm uses the stochastic gradient descent method. The learning rate is initially set to 0.001 and is dynamically adjusted according to the degree of performance improvement. The control strategy parameters are updated every 100 time steps.

[0130] A safety monitoring module is set up to monitor key operating parameters in real time. When an abnormality is detected or a constraint is about to be violated, the preset safety protection strategy is triggered. The safety protection strategy includes multi-level response mechanisms such as rapid load reduction and emergency shutdown. The key parameters monitored include boiler steam temperature and pressure, turbine vibration, bearing temperature, etc. The sampling frequency is 100Hz.

[0131] In the actual application of a 600MW subcritical unit, after adopting the above method, the unit efficiency was improved by 1.8 percentage points, NOx emissions were reduced by 22%, the load change rate was increased from 2% rated load / minute to 6% rated load / minute, and the minimum technical output was reduced from 40% rated load to 22% rated load. At the same time, the expected life of major equipment was extended by about 15%. Under the condition of frequent peak regulation of the power grid, the annual equivalent utilization hours of the unit increased by about 450 hours.

[0132] In this embodiment, multi-objective optimization and deep reinforcement learning are used to significantly improve the load change rate of the unit and reduce the minimum technical output, which greatly enhances the peak-shaving flexibility of the unit and can better adapt to the large-scale grid connection of renewable energy. By incorporating the maximization of equipment life into the optimization objectives and adopting an evaluation method based on a fatigue damage model, the loss of equipment is effectively reduced, which not only reduces maintenance costs but also improves the reliability and availability of equipment. By combining model predictive control and deep reinforcement learning, the control strategy can be adaptively adjusted according to system status and environmental changes, and can better cope with uncertain factors such as load fluctuations and changes in fuel quality, thereby improving the robustness of the system. In summary, this embodiment not only improves the performance of a single unit, but also provides strong support for the flexibility and reliability of the entire power system, which is of great significance to promoting energy transformation and sustainable development.

[0133] Figure 2 Schematic diagram of the structure of the system of the embodiment of the present invention, as shown in Figure 2 As shown, the system comprises:

[0134] The first unit is used to collect all-round operation data of thermal power plants using distributed sensor networks, pre-process the collected data using edge computing technology, including outlier detection, data smoothing and compression, and build a multi-level dynamic nonlinear model of thermal power plants based on the pre-processed data, including fuel delivery system, boiler system, turbine system, generator system and pollutant control system. A hybrid deep learning algorithm is used in combination with long short-term memory network and convolutional neural network to train and optimize the models of each subsystem. The improved adaptive quantum behavior particle swarm optimization algorithm is used to identify the model parameters in real time, generate identification results and establish a multi-dimensional thermal power plant operation characteristic database based on the identification results, including efficiency curves, emission characteristics and dynamic response of start-stop process under different loads;

[0135] The second unit is used to analyze the operating bottlenecks of the units based on the operating characteristics database of the thermal power plant using data mining technology, including load change rate limiting factors and minimum technical output constraints, and design multi-objective and multi-scheme flexibility transformation plans, including the design of the intelligent ash cleaning system for boiler anti-ash accumulation, intelligent optimization of the turbine bypass system, frequency conversion transformation of the water supply system and transformation of the combustion optimization system. Virtual reality and augmented reality technologies are used to visualize and optimize the transformation plan. During the transformation implementation process, the digital twin technology based on edge computing is used to build a high-fidelity digital model of the thermal power plant. Through the transformation, the start-stop speed, load change rate and minimum technical output of the units are improved, and the flexibility transformation results are obtained;

[0136] The third unit is used to construct a multi-objective optimization function based on the dynamic nonlinear model and flexibility transformation results. The objectives include maximizing unit efficiency, minimizing emissions, maximizing peak-shaving capacity and maximizing equipment life. An improved model predictive control algorithm is used in combination with a deep reinforcement learning method to formulate an adaptive deep peak-shaving optimization control strategy. Short-term and medium-term forecasts of grid load are made. Based on the forecast results, a deep deterministic policy gradient algorithm is used to generate an initial control strategy. The initial control strategy is optimized and adjusted online. A hierarchical distributed control architecture is designed to generate a global optimization goal. The global goal is decomposed into local goals of each subsystem, and the interactions between subsystems are coordinated. Precise control of each subsystem is achieved through intelligent actuators to achieve coordinated control of the unit. A real-time performance evaluation mechanism based on big data analysis is established. A sliding time window method is used to continuously evaluate and optimize the control strategy. Based on the evaluation results, an adaptive learning algorithm is used to continuously optimize the control strategy.

[0137] According to a third aspect of the embodiments of the present invention,

[0138] An electronic device is provided, comprising:

[0139] processor;

[0140] a memory for storing processor-executable instructions;

[0141] The processor is configured to call the instructions stored in the memory to execute the aforementioned method.

[0142] According to a fourth aspect of the embodiments of the present invention,

[0143] A computer-readable storage medium is provided, on which computer program instructions are stored. When the computer program instructions are executed by a processor, the aforementioned method is implemented.

[0144] The present invention may be a method, an apparatus, a system and / or a computer program product. The computer program product may include a computer-readable storage medium carrying computer-readable program instructions for executing various aspects of the present invention.

[0145] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for optimizing the control of thermal power plant flexibility transformation and deep peak load regulation, characterized in that: include: A distributed sensor network is used to collect all-round operation data of the thermal power plant, and edge computing technology is used to pre-process the collected data, including outlier detection, data smoothing and compression. Based on the pre-processed data, a multi-level dynamic nonlinear model of the thermal power plant is constructed, including the fuel delivery system, boiler system, turbine system, generator system and pollutant control system. A hybrid deep learning algorithm is used, combined with a long short-term memory network and a convolutional neural network, to train and optimize the models of each subsystem. The improved adaptive quantum behavior particle swarm optimization algorithm is used to identify the model parameters in real time, generate identification results, and based on the identification results, a multi-dimensional thermal power plant operation characteristic database is established, including efficiency curves, emission characteristics and dynamic responses of the start-stop process under different loads; Based on the operation characteristic database of thermal power plants, data mining technology is used to analyze the operation bottlenecks of the units, including load change rate limiting factors and minimum technical output constraints, and multi-objective and multi-scheme flexibility transformation plans are designed, including the design of intelligent ash cleaning systems for boiler dust prevention, intelligent optimization of turbine bypass systems, frequency conversion transformation of water supply systems, and transformation of combustion optimization systems. Virtual reality and augmented reality technologies are used to visualize and optimize the transformation plans. During the transformation implementation process, digital twin technology based on edge computing is used to build a high-fidelity digital model of the thermal power plant. Through the transformation, the start-stop speed, load change rate and minimum technical output of the units are improved, and the flexibility transformation results are obtained; A multi-objective optimization function is constructed based on the dynamic nonlinear model and the results of flexibility transformation. The objectives include maximizing unit efficiency, minimizing emissions, maximizing peak-shaving capacity and maximizing equipment life. An improved model predictive control algorithm is used in combination with a deep reinforcement learning method to formulate an adaptive deep peak-shaving optimization control strategy. Short-term and medium-term forecasts of grid load are made. Based on the prediction results, a deep deterministic policy gradient algorithm is used to generate an initial control strategy. The initial control strategy is optimized and adjusted online. A hierarchical distributed control architecture is designed to generate a global optimization goal. The global goal is decomposed into local goals of each subsystem, and the interactions between subsystems are coordinated. Precise control of each subsystem is achieved through intelligent actuators to achieve coordinated control of the unit. A real-time performance evaluation mechanism based on big data analysis is established. A sliding time window method is used to continuously evaluate and optimize the control strategy. Based on the evaluation results, an adaptive learning algorithm is used to continuously optimize the control strategy.

2. The method according to claim 1, characterized in that A distributed sensor network is used to collect all-round operation data of the thermal power plant, and the edge computing technology is used to pre-process the collected data, including outlier detection, data smoothing and compression. Based on the pre-processed data, a multi-level dynamic nonlinear model of the thermal power plant is constructed, including the fuel delivery system, boiler system, turbine system, generator system and pollutant control system. A hybrid deep learning algorithm is used, combined with long short-term memory network and convolutional neural network, to train and optimize the models of each subsystem. The improved adaptive quantum behavior particle swarm optimization algorithm is used to identify the model parameters in real time, generate identification results and establish a multi-dimensional thermal power plant operation characteristic database based on the identification results, including efficiency curves, emission characteristics and start-stop process dynamic responses under different loads. High-precision sensors are placed on the steam turbines, boilers and generators of thermal power plants to collect real-time operating parameters, and edge computing nodes are deployed on site. The edge computing nodes are equipped with ARM Cortex-A72 processors and 4GB RAM, running a lightweight Linux operating system; Using the edge computing node to preprocess the collected data, the preprocessing includes using an improved local anomaly factor algorithm to perform outlier detection, the improved local anomaly factor algorithm reduces the computational complexity by combining a sliding time window technology, using an adaptive Savitzky-Golay filter to perform data smoothing, the adaptive Savitzky-Golay filter automatically adjusts the filter window size and the polynomial order according to the local characteristics of the data, and using a hybrid compression algorithm based on wavelet transform and deep autoencoder to perform data compression to obtain preprocessed data; Based on the preprocessed data, a multi-level dynamic nonlinear model of a thermal power plant is constructed, the model includes five subsystems: a fuel delivery system, a boiler system, a steam turbine system, a generator system and a pollutant control system; A hybrid deep learning algorithm is used to train and optimize each subsystem model. The hybrid deep learning algorithm combines a long short-term memory network and a convolutional neural network, wherein the long short-term memory network adopts a bidirectional structure, the convolutional neural network adopts a multi-scale convolution kernel, and the outputs of the long short-term memory network and the convolutional neural network are fused through an attention mechanism; An improved adaptive quantum behavior particle swarm optimization algorithm is used to realize real-time identification of model parameters, quantum states are used to represent particle positions and update phase angles, and quantum revolving gates are used to realize particle motion to obtain model identification results. Based on the model identification results, a multi-dimensional thermal power plant operation characteristic database is constructed. The database adopts a distributed storage architecture, uses Apache Cassandra as the underlying storage system, and develops a data access interface based on RESTful API, which supports multi-parameter joint queries and time series slicing. AES-256 is used to encrypt and store data, implement role-based access control, and record audit logs for data access operations.

3. The method according to claim 2, characterized in that Using quantum states to represent particle positions and update phase angles is shown in the following formula: Δθ=ωΔθ prev +c1r1(θ pbest -θ)+c2r2(θ gbest -θ)+α·(rand()-0.5)·π; Among them, △θ is the change of phase angle, which indicates the update amplitude of the current phase angle, ω indicates the inertia weight, and △θ prev represents the phase angle change of the previous step, c1 represents the individual learning factor, r1 represents a random number between 0 and 1, θ pbest represents the individual optimal phase angle, c2 represents the global learning factor, r2 represents another random number between 0 and 1, θ gbest represents the global optimal phase angle, α represents the interference intensity, and rand() represents the uncertainty function, which is used to randomly generate a random number between 0 and 1.

4. The method according to claim 1, characterized in that Based on the operation characteristic database of thermal power plants, data mining technology is used to analyze the operation bottlenecks of the units, including load change rate limiting factors and minimum technical output constraints, and multi-objective and multi-scheme flexibility transformation plans are designed, including the design of boiler anti-ashing intelligent cleaning system, intelligent optimization of turbine bypass system, frequency conversion transformation of water supply system and combustion optimization system transformation. Virtual reality and augmented reality technologies are used to visualize and optimize the transformation plan. During the transformation implementation process, digital twin technology based on edge computing is used to build a high-fidelity digital model of the thermal power plant. The start-stop speed, load change rate and minimum technical output of the units are improved through transformation. The flexibility transformation results include: Constructing a dynamic nonlinear model of the boiler system, and developing an intelligent combustion optimization system based on the dynamic nonlinear model of the boiler system, wherein the intelligent combustion optimization system includes using a long short-term memory network to predict future load changes, using an improved nonlinear model predictive control algorithm to optimize the coal feed rate, primary air volume and secondary air volume in real time, using a multi-objective optimization strategy to construct an optimization objective function, and using a sequential quadratic programming algorithm combined with a trust region method to solve nonlinear optimization problems; Construct a steam turbine system model, and develop an adaptive sliding pressure operation control strategy based on the steam turbine system model, wherein the adaptive sliding pressure operation control strategy includes training the control strategy using a deep deterministic policy gradient algorithm, defining a state space including main steam pressure, temperature, flow rate and generator output power, defining an action space as parameters of a sliding pressure curve and designing a reward function, deploying edge computing nodes on feedwater pumps, induced draft fans and forced draft fans, running a lightweight model predictive control algorithm on each edge computing node, designing a distributed coordination mechanism based on federated learning, wherein each edge node maintains a local model, updates model parameters using local data, and sends the updated model parameters to a central server, wherein the central server aggregates all updates and sends the aggregated updates back to each edge node, and the edge node updates the local model using the received updates; Develop a comprehensive intelligent dispatching system, treat the boiler system, turbine system and auxiliary system as independent agents, adopt an improved multi-agent soft actor-critic algorithm to achieve overall optimal control, each agent has an agent strategy network and value function network corresponding to the current agent, introduce a global value function to evaluate the value of joint actions, and make decisions based on local value functions and global value functions; Visual simulation is carried out to construct a high-fidelity digital model of the thermal power plant. Through transformation, the start-stop speed, load change rate and minimum technical output of the units are improved to obtain flexible transformation results.

5. The method according to claim 4, characterized in that The update target of the value function network is shown in the following formula: Among them, E represents expectation, s t Indicates the current state, a t Indicates the current action, r t represents the current reward value, s t+1 represents the next state, D represents the experience replay buffer, Qi(s t , a t ) represents the value function network for the current state s t and the current action a t The estimated value of , γ represents the discount factor, According to the strategy π i From the next state s t+1 Randomly select action a t+1 The expected value, Q i (s t+1 , a t+1 ) represents the value function network for the next state s t+1和 Next action a t+1 The estimated value of , β represents the policy entropy adjustment factor, and P represents the current policy π i In status t+1 Next select action a t+1 The logarithmic probability of .

6. The method according to claim 1, characterized in that According to the dynamic nonlinear model and the results of flexibility transformation, a multi-objective optimization function is constructed. The objectives include maximizing unit efficiency, minimizing emissions, maximizing peak-shaving capacity, and maximizing equipment life. An improved model predictive control algorithm is used, combined with a deep reinforcement learning method, to formulate an adaptive deep peak-shaving optimization control strategy. Short-term and medium-term forecasts of power grid load are made. Based on the forecast results, a deep deterministic policy gradient algorithm is used to generate an initial control strategy. The initial control strategy is optimized and adjusted online. A hierarchical distributed control architecture is designed to generate a global optimization goal. The global goal is decomposed into local goals of each subsystem, and the interactions between subsystems are coordinated. Accurate control of each subsystem is achieved through intelligent actuators to achieve coordinated control of the unit. A real-time performance evaluation mechanism based on big data analysis is established. A sliding time window method is used to continuously evaluate and optimize the control strategy. Based on the evaluation results, an adaptive learning algorithm is used to continuously optimize the control strategy, including: Based on the dynamic nonlinear model and the flexibility transformation results, a multi-objective optimization function is constructed, wherein the objectives include maximizing unit efficiency, minimizing emissions, maximizing peak load regulation capacity, and maximizing equipment life, and an improved model predictive control algorithm is designed, wherein the improved model predictive control algorithm includes a state estimator, a constraint processor, and an optimization solver; Constructing a deep reinforcement learning framework, the deep reinforcement learning framework includes a deep policy network and a value network, the deep policy network is used to generate control actions, and the value network is used to evaluate state values, and combining the improved model predictive control algorithm with the deep reinforcement learning framework to form an adaptive deep peak-shaving optimization control strategy; A method combining a long short-term memory network and a convolutional neural network is used to perform short-term and medium-term forecasts on the power grid load to obtain load forecast results. Based on the load forecast results, an initial control strategy is generated according to the improved model predictive control algorithm and a deep deterministic policy gradient algorithm. The deep deterministic policy gradient algorithm includes an actor network and a critic network. Both the actor network and the critic network adopt a four-layer fully connected neural network structure. Design an online optimization module to optimize and adjust the initial control strategy in real time. The online optimization module adopts a model predictive control method to solve the optimal control sequence in the rolling time domain and construct a hierarchical distributed control architecture. The hierarchical distributed control architecture includes a global coordination layer and a local control layer. The global coordination layer is responsible for generating the global optimization target. The decomposition and coordination algorithm is used to decompose the global target into the local targets of each subsystem. The local control layer is responsible for executing the precise control of each subsystem. The adaptive PID control algorithm is used to track the local targets. Design an intelligent actuator. The intelligent actuator includes an advanced actuator and an embedded controller to achieve precise control of each subsystem. Establish a real-time performance evaluation mechanism based on big data analysis. Use a sliding time window method to continuously evaluate and optimize the adaptive deep peak-shaving optimization control strategy. Based on the evaluation results of the real-time performance evaluation mechanism, an adaptive learning algorithm is used to continuously optimize the adaptive deep peak-shaving optimization control strategy. The adaptive learning algorithm uses the gradient descent method to update the control strategy parameters. A safety monitoring module is set to monitor key operating parameters in real time. When an anomaly is detected or a constraint is about to be violated, a preset safety protection strategy is triggered.

7. A thermal power plant flexibility transformation and deep peak load optimization control system, used to implement the method described in any one of claims 1 to 6, characterized in that: include: The first unit is used to collect all-round operation data of thermal power plants using distributed sensor networks, pre-process the collected data using edge computing technology, including outlier detection, data smoothing and compression, and build a multi-level dynamic nonlinear model of thermal power plants based on the pre-processed data, including fuel delivery system, boiler system, turbine system, generator system and pollutant control system. A hybrid deep learning algorithm is used in combination with long short-term memory network and convolutional neural network to train and optimize the models of each subsystem. The improved adaptive quantum behavior particle swarm optimization algorithm is used to identify the model parameters in real time, generate identification results and establish a multi-dimensional thermal power plant operation characteristic database based on the identification results, including efficiency curves, emission characteristics and dynamic response of start-stop process under different loads; The second unit is used to analyze the operating bottlenecks of the units based on the operating characteristics database of the thermal power plant using data mining technology, including load change rate limiting factors and minimum technical output constraints, and design multi-objective and multi-scheme flexibility transformation plans, including the design of the intelligent ash cleaning system for boiler anti-ash accumulation, intelligent optimization of the turbine bypass system, frequency conversion transformation of the water supply system and transformation of the combustion optimization system. Virtual reality and augmented reality technologies are used to visualize and optimize the transformation plan. During the transformation implementation process, the digital twin technology based on edge computing is used to build a high-fidelity digital model of the thermal power plant. Through the transformation, the start-stop speed, load change rate and minimum technical output of the units are improved, and the flexibility transformation results are obtained; The third unit is used to construct a multi-objective optimization function based on the dynamic nonlinear model and flexibility transformation results. The objectives include maximizing unit efficiency, minimizing emissions, maximizing peak-shaving capacity and maximizing equipment life. An improved model predictive control algorithm is used in combination with a deep reinforcement learning method to formulate an adaptive deep peak-shaving optimization control strategy. Short-term and medium-term forecasts of grid load are made. Based on the forecast results, a deep deterministic policy gradient algorithm is used to generate an initial control strategy. The initial control strategy is optimized and adjusted online. A hierarchical distributed control architecture is designed to generate a global optimization goal. The global goal is decomposed into local goals of each subsystem, and the interactions between subsystems are coordinated. Precise control of each subsystem is achieved through intelligent actuators to achieve coordinated control of the unit. A real-time performance evaluation mechanism based on big data analysis is established. A sliding time window method is used to continuously evaluate and optimize the control strategy. Based on the evaluation results, an adaptive learning algorithm is used to continuously optimize the control strategy.

8. An electronic device, characterized in that: include: processor; a memory for storing processor-executable instructions; The processor is configured to call the instructions stored in the memory to execute the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having computer program instructions stored thereon, characterized in that: When the computer program instructions are executed by a processor, the method according to any one of claims 1 to 6 is implemented.

Images