Intelligent operation scheduling management system of energy storage power station

Abstract

The invention relates to the technical field of energy storage battery equalization adjustment, in particular to an energy storage power station intelligent operation scheduling management system which comprises a battery parameter estimation unit, a dynamic equalization threshold generation unit, an equalization circuit control unit, a data fusion prediction unit and a scheduling strategy generation unit. The battery parameter estimation unit calculates key parameters of a single battery, the dynamic equalization threshold generation unit generates a dynamic equalization threshold through a fuzzy logic algorithm in combination with a battery aging stage, multiple parameters and temperature factors, and the equalization circuit control unit determines an energy transfer strategy according to a model prediction control algorithm. A switch tube is controlled through an adaptive phase shift angle adjustment technology, a fault diagnosis and fault-tolerant control mechanism is provided, the data fusion prediction unit predicts key indexes of the energy storage power station, the scheduling strategy generation unit formulates an optimal operation scheduling strategy, and the ability of the energy storage power station to participate in power grid auxiliary service is improved.

Description

Technical Field

[0001] This invention relates to the field of energy storage battery balancing and regulation technology, and more specifically, to an intelligent operation and scheduling management system for energy storage power stations. Background Technology

[0002] Battery balancing is an important technology, and battery balancing management is a key aspect of ensuring the performance and lifespan of energy storage systems.

[0003] Traditional battery balancing management typically uses fixed balancing thresholds. However, as batteries undergo continuous changes in state of charge, open-circuit voltage, internal resistance, and aging during use, fixed thresholds lead to excessively frequent balancing operations in the early stages of battery aging, increasing energy loss. In the later stages of aging, the state of charge and open-circuit voltage changes become more complex, and the differences between different batteries widen further. Traditional fixed threshold management does not consider these changes and continues to perform balancing operations according to predetermined thresholds in the later stages of aging. When the differences between batteries within the battery pack exceed the fixed threshold range, the lack of flexibility in threshold setting prevents timely balancing, causing voltage differences to accumulate and resulting in poor voltage consistency within the battery pack, affecting the overall performance of the energy storage power station. Furthermore, battery performance varies under different ambient temperatures, making it difficult for fixed thresholds to guarantee balancing effectiveness. To address this technical problem, we provide an intelligent operation and scheduling management system for energy storage power stations. Summary of the Invention

[0004] The purpose of this invention is to provide an intelligent operation and scheduling management system for energy storage power stations to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, an intelligent operation and scheduling management system for energy storage power stations is provided, including a battery parameter estimation unit, a dynamic balancing threshold generation unit, a balancing circuit control unit, a data fusion and prediction unit, and a scheduling strategy generation unit.

[0006] The battery parameter estimation unit collects the current and voltage data of individual cells, and calculates the state of charge, open-circuit voltage and internal resistance of individual cells by driving the Kalman filter algorithm iteration and introducing the Nernst equation optimization algorithm.

[0007] The dynamic equilibrium threshold generation unit predicts the battery aging stage through the battery life prediction model, and inputs the state of charge, open circuit voltage and internal resistance of the individual cells into the fuzzy logic algorithm. The dynamic equilibrium threshold is then output to the battery management system through reasoning based on the preset fuzzy rule base.

[0008] The equalization circuit control unit receives the equalization command issued by the battery management system, determines the direction and magnitude of energy transfer based on the output of the battery parameter estimation unit and the dynamic equalization threshold generation unit, and controls the switching transistor to turn on and off based on pulse width modulation phase shift control technology.

[0009] The data fusion prediction unit collects meteorological, battery and power grid interaction data and constructs a hybrid model of convolutional neural network and long short-term memory network. The model outputs the future available charging and discharging capacity of the energy storage power station, the optimal charging and discharging time period and the potential to participate in grid ancillary services.

[0010] The scheduling strategy generation unit formulates the optimal operation and scheduling strategy based on the model output and the current power grid demand using optimization algorithms.

[0011] As a further improvement to this technical solution, the battery parameter estimation unit introduces a particle filter auxiliary mechanism during the iteration process of driving the Kalman filter algorithm. It monitors the data residuals of the Kalman filter algorithm in real time. When the standard deviation of the data residuals exceeds the preset residual threshold, the particle filter algorithm is started to generate a certain number of particles to simulate the probability distribution of the battery state. Using the particle weight update and resampling steps, combined with the existing iteration results of the Kalman filter, the state of charge, open circuit voltage and internal resistance of the individual battery are calculated.

[0012] As a further improvement to this technical solution, the dynamic equalization threshold generation unit, when inputting the state of charge, open-circuit voltage, internal resistance, and battery aging stage of a single battery cell into the fuzzy logic algorithm, adopts a multimodal feature fusion method. The state of charge, open-circuit voltage, internal resistance, and battery aging stage are mapped to different feature spaces respectively. By constructing a multimodal fusion network, these features are fused. This network consists of multiple fully connected layers and activation functions, and a batch normalization layer is added after each fully connected layer. The fused features are then input into the fuzzy logic algorithm.

[0013] As a further improvement to this technical solution, the preset fuzzy rule base of the dynamic equilibrium threshold generation unit adopts a fuzzy rule generation method based on genetic algorithm optimization. Initially, a set of fuzzy rule populations are randomly generated. Each rule consists of preconditions and conclusions, and a fitness function is defined. This function considers the voltage consistency and energy loss of the battery pack. The rule population is iteratively updated through genetic operations of selection, crossover, and mutation until an optimal set of fuzzy rules is found and used as the preset fuzzy rule base.

[0014] As a further improvement to this technical solution, the dynamic equilibrium threshold generation unit also considers the influence of battery temperature on the dynamic equilibrium threshold. Temperature is introduced as a new input variable in the fuzzy logic algorithm. Based on historical experience, a correlation model is established between temperature and state of charge, open circuit voltage, internal resistance, and battery aging stage. The weights of rules in the fuzzy rule base are adjusted according to the model, and the dynamic equilibrium threshold is output.

[0015] As a further improvement to this technical solution, when determining the direction and magnitude of energy transfer, the equalization circuit control unit uses a model predictive control-based energy distribution optimization algorithm to establish a dynamic model of the battery pack. Based on the results output by the battery parameter estimation unit and the dynamic equalization threshold generation unit, it predicts the state changes of the battery pack in the future. With the goal of minimizing the energy loss of the battery pack and maximizing the equalization effect of the battery pack, it solves the optimal energy transfer strategy in the prediction time domain and updates the control strategy through rolling optimization.

[0016] As a further improvement to this technical solution, the equalization circuit control unit uses pulse width modulation phase shift control technology to control the switching transistor to turn on and off. It adopts an adaptive phase shift angle adjustment method to monitor the voltage and current changes of the battery pack in real time. It dynamically adjusts the phase shift angle of the pulse according to the energy transfer requirements and the real-time status of the battery pack. By establishing a mapping relationship between the phase shift angle and the energy transfer efficiency, it uses a reinforcement learning algorithm to select the optimal phase shift angle based on the current energy transfer effect and battery status.

[0017] As a further improvement to this technical solution, the equalization circuit control unit also considers the loss factor of the equalization circuit. When determining the energy transfer direction and magnitude, a loss model of the equalization circuit is established. When optimizing the energy transfer strategy, the loss of the equalization circuit is used as a constraint condition and a segmented energy transfer method is adopted for energy transfer.

[0018] As a further improvement to this technical solution, the equalization circuit control unit adopts a fault diagnosis and fault-tolerant control mechanism. During the operation of the equalization circuit, it monitors the current, voltage and temperature parameters of the switching transistor in real time, and determines whether the switching transistor has failed through fault feature extraction and pattern recognition algorithms. Once a fault is detected, the fault-tolerant control strategy is immediately activated.

[0019] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0020] In the intelligent operation and dispatch management system for energy storage power stations, the dynamic balancing threshold generation unit predicts the aging stage of the battery using a battery life prediction model. Combining the state of charge, open-circuit voltage, and internal resistance parameters, it outputs a dynamic balancing threshold using a fuzzy logic algorithm. It also employs a multi-modal feature fusion method and a fuzzy rule generation method based on genetic algorithm optimization, and has an online learning mechanism to make the threshold more closely match the actual situation of the battery, effectively improving the voltage consistency of the battery pack, reducing energy loss, and extending battery life. After receiving the balancing command, the balancing circuit control unit uses an energy distribution optimization algorithm based on model predictive control to determine the direction and magnitude of energy transfer. It optimizes the energy transfer efficiency through an adaptive phase shift angle adjustment method, and also considers the loss factors of the balancing circuit and adopts a segmented energy transfer method. At the same time, it has fault diagnosis and fault-tolerant control mechanisms to ensure the stable and reliable operation of the balancing circuit and ensure the efficient, safe, and continuous operation of the energy storage power station. Attached Figure Description

[0021] Figure 1 This is an overall block diagram of the present invention.

[0022] The meanings of the labels in the diagram are as follows:

[0023] 1. Battery parameter estimation unit; 2. Dynamic equalization threshold generation unit; 3. Equalization circuit control unit; 4. Data fusion prediction unit; 5. Scheduling strategy generation unit. Detailed Implementation

[0024] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0025] This invention provides an intelligent operation and scheduling management system for energy storage power stations. Please refer to [link / reference]. Figure 1 As shown, it includes a battery parameter estimation unit 1, a dynamic equalization threshold generation unit 2, an equalization circuit control unit 3, a data fusion prediction unit 4, and a scheduling strategy generation unit 5;

[0026] The battery parameter estimation unit 1 collects the current and voltage data of individual cells, and calculates the state of charge, open-circuit voltage and internal resistance of individual cells by driving the Kalman filter algorithm iteration and introducing the Nernst equation optimization algorithm.

[0027] The battery parameter estimation unit 1 introduces a particle filter auxiliary mechanism during the iteration of the driving Kalman filter algorithm. This mechanism monitors the data residuals of the Kalman filter algorithm in real time. The data residuals directly reflect the deviation between the predicted values ​​and actual measured values ​​of the Kalman filter model. A preset residual threshold is set. When the standard deviation of the data residuals exceeds the preset residual threshold, the particle filter algorithm is activated to generate a certain number of particles to simulate the probability distribution of the battery state. The generation of particles is based on prior knowledge of the battery state. Each particle represents a possible battery state, including dimensions such as state of charge, open-circuit voltage, and internal resistance parameters, increasing the probability of finding more accurate battery parameters. The generated particles initially have the same weight, which cannot accurately reflect their fit with the actual battery state. Through weight updates, weights are allocated according to the closeness between the predicted state represented by the particle and the actual measured value, resulting in particles that are closer to the real situation. The particle weight update and resampling steps improve the long-term reliability of the particle filter algorithm, continuously providing high-quality samples for battery parameter estimation. Combined with existing iteration results of the Kalman filter, the state of charge, open-circuit voltage, and internal resistance of individual batteries are calculated, providing a data foundation for the subsequent operation and management of the energy storage power station.

[0028] The dynamic equilibrium threshold generation unit 2 predicts the battery aging stage through the battery life prediction model, and inputs the state of charge, open circuit voltage and internal resistance of the individual cells into the fuzzy logic algorithm. The dynamic equilibrium threshold is then output to the battery management system through reasoning based on the preset fuzzy rule base.

[0029] The dynamic equilibrium threshold generation unit 2 employs a multimodal feature fusion method when inputting the state of charge, open-circuit voltage, internal resistance, and battery aging stage of a single battery cell into the fuzzy logic algorithm. This method maps the state of charge, open-circuit voltage, internal resistance, and battery aging stage to different feature spaces, allowing the model to more specifically capture the unique information of each feature and avoid mutual interference between features. This prepares the model for subsequent accurate fusion. By constructing a multimodal fusion network, these features are fused, enabling the network to adaptively learn the deep patterns inherent in each feature combination. This provides the fuzzy logic algorithm with more representative comprehensive features. The network consists of multiple fully connected layers and activation functions, with each fully connected layer followed by... Adding a batch normalization layer makes the training of the multimodal fusion network more efficient and stable, resulting in higher quality fused features and stronger adaptability to different battery operating conditions. The fused features are then input into a fuzzy logic algorithm. The fuzzy logic algorithm first fuzzifies the input features, defines fuzzy sets ("low charge", "medium charge", "high charge" fuzzy sets and corresponding membership functions for the state of charge), determines the degree to which each feature value belongs to each fuzzy set, and then performs fuzzy inference based on a preset fuzzy rule library. Through fuzzy synthesis and defuzzification operations, it finally outputs a dynamic equalization threshold, providing reasonable and precise control commands for the battery equalization circuit, extending battery life, and improving the operational stability and efficiency of the energy storage power station.

[0030] The preset fuzzy rule base of the dynamic equilibrium threshold generation unit 2 adopts a fuzzy rule generation method based on genetic algorithm optimization. Since the relationship between battery state and equilibrium threshold is complex, it is difficult to directly determine the optimal fuzzy rule. By randomly generating a set of fuzzy rule populations, it can broadly cover various possible rule combinations. Initially, a set of fuzzy rule populations is randomly generated. Each rule consists of a premise and a conclusion. The form of the fuzzy rule is set as "IF (premise) THEN (conclusion)". The premise involves the state of charge (SOC), open-circuit voltage (OCV), internal resistance (R), and battery aging stage (LA) of a single cell. Each variable is divided into several fuzzy subsets, which are randomly generated. The initial population is composed of a set of fuzzy rules. For example, a rule could be: "IFSOC is moderate, ANDOCV is high, ANDR is normal, ANDLA is early, THEN is moderate." The conclusion of each rule is also a fuzzy description of the equilibrium threshold, corresponding to different value ranges, increasing the probability of finding high-quality rules, and defining a fitness function. This function considers the voltage consistency and energy loss of the battery pack, and iteratively updates the rule population through genetic operations of selection, crossover, and mutation. The selection operation aims to select individuals with high fitness from the current population, giving them more opportunities to participate in subsequent genetic operations, gradually increasing the proportion of high-quality rules in the population, and developing towards a better equilibrium strategy, thereby improving the battery's equilibrium effect. The crossover operation simulates biological gene recombination, creating new offspring individuals by exchanging partial gene fragments of two parent individuals, discovering new effective equilibrium strategies, raising the upper limit of battery equilibrium performance, and introducing new rule combination patterns. The mutation operation is equivalent to gene mutation in biological evolution, introducing random changes to the population, avoiding the algorithm from converging to a local optimum too early. Through this random perturbation, a set of optimal fuzzy rules that highly adapt to the battery equilibrium requirements is finally found, and this is used as a preset fuzzy rule library for subsequent dynamic equilibrium threshold inference output, thereby improving the equilibrium performance of the battery management system.

[0031] The dynamic equilibrium threshold generation unit 2 also considers the impact of battery temperature on the dynamic equilibrium threshold. Temperature is introduced as a new input variable into the fuzzy logic algorithm, which can more comprehensively reflect the actual operating conditions of the battery, making the generated dynamic equilibrium threshold closely match the real-time state of the battery. Based on historical experience, a correlation model is established between temperature and state of charge, open-circuit voltage, internal resistance, and battery aging stages. Temperature and various key battery performance indicators have an inherent correlation, but the relationship is complex and non-linear. By establishing a correlation model, these relationships are quantified, providing a scientific basis for subsequent precise adjustment of fuzzy rule weights based on temperature. The weights of rules in the fuzzy rule library are adjusted according to this model. The original rule weights in the fuzzy rule library are set based on general operating conditions. After adding the temperature variable, the weights need to be dynamically adjusted based on the correlation between temperature and other variables. The correlation model is used to calculate the weights when... The algorithm analyzes the changing trends and mutual influence of each input variable at the previous temperature. Based on these trends and influence, the weights are adjusted to improve the battery's balancing efficiency under complex temperature conditions. Through temperature introduction, correlation modeling, and weight adjustment, the fuzzy logic algorithm integrates comprehensive information. Then, based on optimized rules, a dynamic balancing threshold is output to guide battery balancing operations, achieving precise battery energy management. The algorithm inputs an input vector containing temperature information into the fuzzy logic algorithm. First, each input variable is fuzzified. Membership degrees are determined based on preset fuzzy sets and membership functions. Then, based on the adjusted fuzzy rule base, fuzzy inference is performed through fuzzy synthesis and defuzzification operations, ultimately outputting a dynamic balancing threshold. This ensures the battery pack maintains a good balancing state under different temperature conditions, extending battery life and improving the overall performance of the energy storage power station.

[0032] The equalization circuit control unit 3 receives the equalization command issued by the battery management system, determines the direction and magnitude of energy transfer based on the output of the battery parameter estimation unit 1 and the dynamic equalization threshold generation unit 2, and controls the switching transistor to turn on and off based on pulse width modulation phase shift control technology.

[0033] When determining the direction and magnitude of energy transfer, the equalization circuit control unit 3 uses a model predictive control-based energy distribution optimization algorithm to establish a dynamic model of the battery pack. This model describes the changes in parameters such as voltage, current, and state of charge of the battery pack over time, providing a foundation for subsequent energy distribution optimization. Based on the outputs of the battery parameter estimation unit 1 and the dynamic equalization threshold generation unit 2, it predicts the state changes of the battery pack over a future period, enabling early understanding of the battery pack's development trend and providing more information for energy distribution optimization. This allows for the formulation of a more reasonable control strategy. To minimize energy loss and maximize the equalization effect of the battery pack, an objective function needs to be defined and quantified to solve for the optimal energy transfer strategy in the prediction time domain. The objective function comprehensively considers both the equalization effect and energy loss objectives. The weighting coefficients can be flexibly adjusted to balance these two objectives according to actual needs, making the control strategy more in line with the requirements of practical applications. Solving for the minimum value of the objective function in the prediction time domain, i.e. the optimal energy transfer strategy, enables the battery pack to achieve the best balance and minimum energy loss in the future. Since the actual operation of the battery pack is constantly changing and the prediction model also has certain errors, a rolling optimization method is required to continuously update the control strategy. That is, in each sampling period, the state prediction and objective function are re-solved based on the current battery pack state information to obtain a new optimal control input sequence. Then, only the first control input of the sequence is executed, and the above process is repeated in the next sampling period. This ensures the effectiveness and stability of the control strategy throughout the entire operation, so that the battery pack can always maintain a good balance and low energy loss.

[0034] The equalization circuit control unit 3, based on pulse width modulation (PWM) phase-shift control technology, controls the switching transistor's on / off states. By installing voltage and current sensors in the equalization circuit, it samples the battery pack's voltage and current at a specific sampling frequency, monitoring their changes in real time. This provides accurate data for subsequent dynamic adjustment of the pulse phase-shift angle. Different battery pack states and energy transfer requirements necessitate different phase-shift angles to achieve optimal energy transfer. The required energy transfer amount is calculated based on the monitored battery pack voltage and current and the energy transfer target. To more accurately select the optimal phase-shift angle, the relationship between the phase-shift angle and energy transfer efficiency needs to be clarified. Then, based on preset empirical rules or preliminary mapping relationships, a preliminary phase-shift angle adjustment is determined, improving the equalization circuit's adaptability. By collecting battery pack voltage and current data at different phase-shift angles, the corresponding energy transfer efficiency is calculated, quantifying the complex relationship between the phase-shift angle and energy transfer efficiency. This provides a scientific basis for selecting the optimal phase-shift angle and avoids... Blindly adjusting the phase shift angle leads to efficiency loss. Finally, a reinforcement learning algorithm is used to select the optimal phase shift angle based on the current energy transfer effect and battery state. The reinforcement learning algorithm has adaptive and learning capabilities, continuously exploring and learning the optimal phase shift angle selection strategy based on the current energy transfer effect and battery state, thereby maximizing energy transfer efficiency. A state space, action space, and reward function are defined. The state space includes the battery pack's voltage, current, current phase shift angle, and energy transfer efficiency. The action space is the adjustment range of the phase shift angle. The reward function is used to evaluate the merits of each action, aiming to encourage improved energy transfer efficiency while avoiding excessive phase shift angle adjustments. At each sampling time, an action is selected based on the current state, and the action is executed to obtain the state and reward for the next time step. The reinforcement learning algorithm is then updated with the state and reward for the next time step until it outputs the optimal phase shift angle under various battery pack operating conditions, maximizing energy transfer efficiency and improving the performance and reliability of the equalization circuit.

[0035] The balancing circuit incurs losses during energy transfer, which affect the overall energy efficiency of the battery pack. The balancing circuit control unit 3 also considers the loss factors of the balancing circuit and establishes a loss model of the balancing circuit when determining the direction and magnitude of energy transfer. This model can accurately quantify the energy loss of the balancing circuit under different operating states, providing a basis for subsequent optimization of the energy transfer strategy. The losses of the balancing circuit mainly include the conduction loss of the switching transistor, the switching loss of the switching transistor, the inductor loss, and the capacitor loss. Quantifying the losses of the balancing circuit provides accurate loss information for the optimization of the energy transfer strategy, which helps to formulate a more reasonable control strategy. When optimizing the energy transfer strategy, it is necessary to minimize the losses of the balancing circuit while ensuring the balancing effect of the battery pack. Using losses as constraints makes the optimization process more in line with actual needs, avoiding excessive energy loss due to the pursuit of rapid equalization. In the energy distribution optimization algorithm based on model predictive control, the loss of the equalization circuit is added as a constraint to the objective function, and an upper limit of loss is set as a constraint. While ensuring the equalization effect of the battery pack, the energy loss of the equalization circuit is effectively reduced, improving the overall performance and economy of the battery pack. A segmented energy transfer method is adopted for energy transfer. The segmented energy transfer method can divide the energy transfer process into multiple stages according to the real-time status of the battery pack and the equalization requirements. Each stage uses a different transfer power, thereby achieving low losses in different stages, further reducing the energy loss of the equalization circuit, while accelerating the equalization speed of the battery pack, improving the consistency and service life of the battery pack.

[0036] The equalization circuit control unit 3 adopts a fault diagnosis and fault-tolerant control mechanism. During the operation of the equalization circuit, it monitors the current, voltage, and temperature parameters of the switching transistors in real time, obtains the operating information of the switching transistors in a timely manner, and provides a data foundation for subsequent fault diagnosis. It determines whether the switching transistors have failed through fault feature extraction and pattern recognition algorithms, calculates the statistical characteristics such as the mean, variance, maximum, and minimum values ​​of current, voltage, and temperature, performs Fourier transform on the collected signals to obtain their spectral characteristics, and extracts the amplitude and phase features of specific frequency components, so that the fault diagnosis is more focused on key information and can more accurately identify the fault type and degree. Support vector machine is used as the pattern recognition algorithm. First, it collects the current, voltage, and temperature feature data of the switching transistors under normal operating conditions and various fault conditions to form a training sample set. The training sample set is divided into feature vectors and corresponding labels. The label for normal state is 0, and the labels for different fault states are 1, 2, ... Then, the support vector machine algorithm is used to train the training samples to obtain a classification model. In practical applications, the fault features extracted in real time are input into the trained classification model. The model output determines whether the switching transistor has failed and the type of failure, preventing the failure from escalating further. Once a fault is detected, a fault-tolerant control strategy is immediately activated. A backup switching transistor is pre-set in the equalization circuit. When a fault is detected in a switching transistor, the control unit immediately issues a command to disconnect the faulty switching transistor and connect the backup switching transistor to the circuit to replace the faulty switching transistor and continue working. If there is no backup switching transistor or the number of backup switching transistors is insufficient, the energy transfer path is adjusted according to the location and function of the faulty switching transistor, thereby improving the reliability and safety of the entire battery management system.

[0037] The data fusion prediction unit 4 collects meteorological, battery and grid interaction data and constructs a hybrid model of convolutional neural network and long short-term memory network. The model outputs the future available charging and discharging capacity of the energy storage power station, the optimal charging and discharging time period and the potential to participate in grid ancillary services.

[0038] The scheduling strategy generation unit 5 formulates the optimal operation scheduling strategy based on the model output and the current power grid demand using optimization algorithms.

[0039] In this invention, the battery parameter estimation unit 1 calculates the key parameters of a single battery cell, the dynamic equilibrium threshold generation unit 2 combines the battery aging stage, multiple parameters and temperature factors, and generates a dynamic equilibrium threshold through a fuzzy logic algorithm, which can be optimized online, the equilibrium circuit control unit 3 determines the energy transfer strategy based on the model predictive control algorithm, controls the switching transistor through adaptive phase angle adjustment technology, and also has a fault diagnosis and fault-tolerant control mechanism, the data fusion prediction unit 4 predicts the key indicators of the energy storage power station, and the scheduling strategy generation unit 5 formulates the optimal operation and scheduling strategy to improve the ability of the energy storage power station to participate in grid ancillary services.

[0040] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely preferred examples and are not intended to limit the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. An intelligent operation and dispatch management system for energy storage power stations, characterized in that, It includes a battery parameter estimation unit (1), a dynamic equalization threshold generation unit (2), an equalization circuit control unit (3), a data fusion prediction unit (4), and a scheduling strategy generation unit (5). The battery parameter estimation unit (1) collects the current and voltage data of a single battery cell, and calculates the state of charge, open-circuit voltage and internal resistance of the single battery cell by driving the Kalman filter algorithm iteration and introducing the Nernst equation optimization algorithm. The dynamic equilibrium threshold generation unit (2) predicts the battery aging stage through the battery life prediction model, and inputs the state of charge, open circuit voltage and internal resistance of the individual battery into the fuzzy logic algorithm, and outputs the dynamic equilibrium threshold to the battery management system through the inference of the preset fuzzy rule base. The equalization circuit control unit (3) receives the equalization command issued by the battery management system, determines the energy transfer direction and magnitude based on the output of the battery parameter estimation unit (1) and the dynamic equalization threshold generation unit (2), and controls the switching tube to turn on and off based on the pulse width modulation phase shift control technology.

2. The intelligent operation and dispatch management system for energy storage power stations according to claim 1, characterized in that, The battery parameter estimation unit (1) introduces a particle filter auxiliary mechanism during the iteration of the driving Kalman filter algorithm, monitors the data residual of the Kalman filter algorithm in real time, and starts the particle filter algorithm to generate a certain number of particles to simulate the probability distribution of the battery state when the standard deviation of the data residual exceeds the preset residual threshold. Using the particle weight update and resampling steps, combined with the existing iteration results of the Kalman filter, the state of charge, open circuit voltage and internal resistance of the single battery are calculated.

3. The intelligent operation and dispatch management system for energy storage power stations according to claim 1, characterized in that, When the dynamic equilibrium threshold generation unit (2) inputs the state of charge, open circuit voltage, internal resistance and battery aging stage of a single cell into the fuzzy logic algorithm, it adopts a multimodal feature fusion method to map the state of charge, open circuit voltage, internal resistance and battery aging stage to different feature spaces. By constructing a multimodal fusion network, these features are fused. The network consists of multiple fully connected layers and activation functions, and a batch normalization layer is added after each fully connected layer. The fused features are then input into the fuzzy logic algorithm.

4. The intelligent operation and dispatch management system for energy storage power stations according to claim 1, characterized in that, The preset fuzzy rule library of the dynamic equilibrium threshold generation unit (2) adopts a fuzzy rule generation method based on genetic algorithm optimization. Initially, a set of fuzzy rule populations are randomly generated. Each rule consists of a premise and a conclusion, and a fitness function is defined. This function considers the voltage consistency and energy loss of the battery pack. The rule population is iteratively updated through genetic operations of selection, crossover and mutation until an optimal set of fuzzy rules is found and used as the preset fuzzy rule library.

5. The intelligent operation and dispatch management system for energy storage power stations according to claim 1, characterized in that, The dynamic equilibrium threshold generation unit (2) also considers the influence of battery temperature on the dynamic equilibrium threshold. Temperature is introduced as a new input variable in the fuzzy logic algorithm. Based on historical experience, a correlation model is established between temperature and state of charge, open circuit voltage, internal resistance and battery aging stage. The weights of rules in the fuzzy rule base are adjusted according to the model, and the dynamic equilibrium threshold is output.

6. The intelligent operation and dispatch management system for energy storage power stations according to claim 1, characterized in that, When determining the direction and magnitude of energy transfer, the equalization circuit control unit (3) uses a model-based predictive control energy distribution optimization algorithm to establish a dynamic model of the battery pack. Based on the results output by the battery parameter estimation unit (1) and the dynamic equalization threshold generation unit (2), it predicts the state changes of the battery pack. With the goal of minimizing the energy loss of the battery pack and maximizing the equalization effect of the battery pack, it solves the optimal energy transfer strategy in the prediction time domain and updates the control strategy through rolling optimization.

7. The intelligent operation and dispatch management system for energy storage power stations according to claim 1, characterized in that, The equalization circuit control unit (3) uses pulse width modulation phase shift control technology to control the switching tube to turn on and off. It adopts an adaptive phase shift angle adjustment method to monitor the voltage and current changes of the battery pack in real time. It dynamically adjusts the phase shift angle of the pulse according to the energy transfer requirements and the real-time status of the battery pack. By establishing a mapping relationship between the phase shift angle and the energy transfer efficiency, it uses a reinforcement learning algorithm to select the optimal phase shift angle based on the current energy transfer effect and battery status.

8. The intelligent operation and dispatch management system for energy storage power stations according to claim 1, characterized in that, The equalization circuit control unit (3) also considers the loss factors of the equalization circuit. When judging the direction and magnitude of energy transfer, it establishes a loss model of the equalization circuit. When optimizing the energy transfer strategy, it takes the loss of the equalization circuit as a constraint and adopts a segmented energy transfer method for energy transfer.

9. The intelligent operation and dispatch management system for energy storage power stations according to claim 1, characterized in that, The equalization circuit control unit (3) adopts a fault diagnosis and fault-tolerant control mechanism. During the operation of the equalization circuit, it monitors the current, voltage and temperature parameters of the switching transistor in real time. It uses fault feature extraction and pattern recognition algorithms to determine whether the switching transistor has failed. Once a fault is detected, the fault-tolerant control strategy is immediately activated.

Images