Distributed photovoltaic self-consistent energy supply system for expressway

Abstract

The invention relates to the technical field of photovoltaic energy supply, in particular to a distributed photovoltaic self-consistent energy supply system for an expressway. The system comprises an analysis and prediction module, a strategy making module, a fault diagnosis module and an optimization scheduling module. The photovoltaic power generation power, the energy storage charge and the load operation power are predicted through the analysis and prediction module, the corresponding control strategy is formulated by using the strategy formulation module, and the unified intelligent coordination control is performed on the photovoltaic power generation, the battery energy storage, the load power utilization load and the power grid according to the prediction result and the control strategy. The power supply proportion of battery energy storage, load electrical load and a power grid can be dynamically adjusted according to the predicted photovoltaic power generation power, the power generation power of a photovoltaic array, the charge state of an energy storage system and the power of the electrical load are monitored in real time by using a fault diagnosis module, and the power supply proportion of the power grid is determined according to a neural network model for fault determination. And the fault type and the fault probability are determined, and fault early warning and diagnosis are carried out.

Description

Technical Field

[0001] The present invention relates to the technical field of photovoltaic energy supply, and in particular to a distributed photovoltaic self-consistent energy supply system for a highway. Background Art

[0002] Focusing on the climate characteristics of Xinjiang and the specific energy supply scenarios of high-speed facilities, we will carry out research on the coupling and integration of green energy and transportation energy supply in accordance with local conditions, and explore the complementary smart microgrid model of source, network, load and storage in transportation scenarios; combined with the current development of information technology and the Internet of Things, we will carry out research and application of new models such as intelligent control, cloud data, and smart regulation to address the pain points of Xinjiang's transportation facilities such as long distances and high operation and maintenance costs.

[0003] Since most of the stations along the highways in Xinjiang are located in uninhabited areas, and some facilities along the highways are rich in resources such as wind energy and solar energy, but the proportion of clean energy utilization is low, in order to achieve unified intelligent coordinated control of photovoltaic power generation, battery energy storage, load power consumption and power grid, improve the utilization rate of clean energy, and at the same time, when a single microgrid branch fails, it can reasonably dispatch excess electricity from nearby branches, reduce dependence on the large power grid, and realize local energy consumption and complementarity. Therefore, we propose a distributed photovoltaic self-consistent energy supply system for highways. Summary of the invention

[0004] The purpose of the present invention is to solve the problem of low light energy utilization rate at various stations along the Xinjiang highway, and the inability to efficiently regulate the power of photovoltaic power generation, battery energy storage, load power consumption and power grid. In order to be able to dynamically adjust the power supply ratio of battery energy storage, load power consumption and power grid according to the predicted photovoltaic power generation power, at the same time, in order to be able to reasonably dispatch the electric energy of the microgrid support points, the local consumption and complementarity of electric energy can be realized.

[0005] To achieve the above-mentioned purpose, the present invention provides a highway distributed photovoltaic self-consistent energy supply system, including an analysis and prediction module, a strategy formulation module, a fault diagnosis module and an optimization scheduling module; The analysis and prediction module collects photovoltaic power generation data, energy storage data and load data through big data collection technology, extracts features from the collected data, collects meteorological data such as solar radiation intensity and battery temperature, uses physical model method to predict photovoltaic power generation, uses battery model method to predict the remaining available capacity of energy storage system and the available charging and discharging power, and uses regression model based on traffic flow and time factors to predict short-term power load fluctuations, and transmits the prediction results to the strategy formulation module; The strategy formulation module formulates the photovoltaic power generation control strategy according to the prediction results of the analysis and prediction module and the maximum power point tracking rule, determines the energy storage system control strategy through the charge state control rule, determines the power load control strategy using the key load protection rule, and optimizes the control strategy using the fuzzy algorithm; The fault diagnosis module monitors the power generation of the photovoltaic array, the charge state of the energy storage system and the power of the power load in real time, establishes a neural network model for fault judgment based on the historical data collected by the analysis and prediction module using big data collection technology, and trains the neural network model, and then performs fault warning and diagnosis based on the output fault category and fault probability; The optimization scheduling module rationally allocates excess power from nearby microgrid supports to meet the power demand of the site based on the fault type determined by the fault diagnosis module, using the network-forming microgrid support function, the bilateral service area centralized control function, and the mutual assistance function.

[0006] As a further improvement of the technical solution, the analysis and prediction module includes a data acquisition unit and a model prediction unit; The data acquisition unit collects the power generation power, voltage, current, light intensity, and temperature of the photovoltaic system, the charge state, charge and discharge current, and temperature of the energy storage system, the power, voltage, and frequency of the grid access, and the power demand and power factor of the power load, and performs feature extraction on the collected data; The model prediction unit predicts photovoltaic power generation using a physical model method, predicts the remaining available capacity of the energy storage system and the available charging and discharging power using a battery model method, and predicts short-term power load fluctuations using a regression model based on traffic flow and time factors.

[0007] As a further improvement of the technical solution, the model prediction unit predicts the photovoltaic power generation power by a physical model method, and the physical model formula is:

[0008] in, is the power generation, is the photoelectric conversion efficiency of the photocell, is the area of ​​the photovoltaic panel, is the solar radiation intensity, is the temperature coefficient, is the actual temperature of the photovoltaic panel, is the reference temperature.

[0009] As a further improvement of the technical solution, the model prediction unit predicts the remaining available capacity and the available charging and discharging power of the energy storage system by a battery model method, and the differential equation formula is:

[0010] in, is the terminal voltage of the battery, is the open circuit voltage of the battery, is the internal resistance, is the polarization capacitance, For current.

[0011] As a further improvement of the technical solution, the strategy formulation module includes a rule determination unit and an optimization algorithm unit; The rule determination unit formulates a photovoltaic power generation control strategy according to the maximum power point tracking rule, determines the energy storage system control strategy through the charge state control rule, and determines the power load control strategy using the key load protection rule; The optimization algorithm unit takes light intensity and temperature as input, and after fuzzification, adjusts the working point of the photovoltaic inverter according to predefined fuzzy rules, formulates the energy storage system control strategy with the state of charge and power load demand as input, and sets the grid interaction rules with the difference between local power generation power and power load and grid access status as input.

[0012] As a further improvement of the technical solution, the rule determination unit installs a high-precision power sensor at the grid access point to monitor the power flow in real time, and formulates an anti-backflow rule by setting a backflow power threshold and a time threshold.

[0013] As a further improvement of the present technical solution, the optimization algorithm unit uses a decision tree algorithm to finally formulate the optimized control strategy.

[0014] As a further improvement of the technical solution, the fault diagnosis module includes a data monitoring unit and a fault determination unit; The data monitoring unit collects data covering various system operating states, including data of normal operating states and known fault states, and monitors the power generation power of the photovoltaic array, the charge state of the energy storage system, and the power of the power load in real time; The fault determination unit uses the fault category and fault probability of the neural network model of fault determination to perform fault warning and diagnosis.

[0015] As a further improvement of the technical solution, when the fault judgment unit establishes the neural network model for fault judgment, a multi-classification cross entropy loss function is used to measure the difference between the model prediction probability distribution and the true probability distribution, and the neural network model is trained with the minimum loss function as the goal, wherein the calculation formula of the multi-classification cross entropy loss function is:

[0016] in, is the probability distribution of the model output, is the probability distribution of the true label, is the number of categories.

[0017] As a further improvement of the present technical solution, when the fault judgment unit establishes a neural network model for fault judgment, it uses the Adam optimization algorithm to combine the advantages of the momentum method and the adaptive learning rate to calculate the first-order moment estimate and the second-order moment estimate of the gradient, and then adjusts the learning rate and updates the parameters based on these two estimates.

[0018] Compared with the prior art, the present invention has the following beneficial effects: The highway distributed photovoltaic self-consistent energy supply system predicts photovoltaic power generation, energy storage charge and load operating power through the analysis and prediction module, formulates the corresponding control strategy using the strategy formulation module, and performs unified intelligent coordinated control of photovoltaic power generation, battery energy storage, load power consumption and power grid according to the prediction results and the control strategy, so as to dynamically adjust the power supply ratio of battery energy storage, load power consumption and power grid according to the predicted photovoltaic power generation; The fault diagnosis module is used to monitor the power generation of the photovoltaic array, the charge state of the energy storage system and the power of the power load in real time. The fault type and fault probability are determined according to the neural network model of fault judgment, and fault warning and diagnosis are carried out. The mesh-type microgrid support function, the bilateral service area centralized control function and the mutual assistance function are used to reasonably allocate the excess electric energy of the nearby microgrid support points to meet the power demand of the site and realize the on-site consumption and complementarity of electric energy. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 It is a schematic diagram of the overall structure flow of the present invention; Figure 2 It is a schematic diagram of the overall details of the present invention; Figure 3 It is a schematic diagram of the fault diagnosis module flow of the present invention.

[0020] The meaning of each number in the figure is: 100. Analysis and prediction module; 110. Data acquisition unit; 120. Model prediction unit; 200. Strategy formulation module; 210. Rule determination unit; 220. Optimization algorithm unit; 300. Fault diagnosis module; 310. Data monitoring unit; 320. Fault determination unit; 400. Optimization scheduling unit. DETAILED DESCRIPTION

[0021] The following will be combined with the accompanying drawings in the present invention to clearly and completely describe the technical solutions 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.

[0022] At present, the utilization rate of light energy at various stations along the Xinjiang highway is low, and it is impossible to efficiently regulate the power of photovoltaic power generation, battery energy storage, load power consumption and power grid. In order to be able to dynamically adjust the power supply ratio of battery energy storage, load power consumption and power grid according to the predicted photovoltaic power generation power, at the same time, in order to be able to reasonably dispatch the electricity of microgrid fulcrums, realize the local consumption and complementarity of electricity.

[0023] Therefore, the present invention predicts photovoltaic power generation, energy storage charge and load operating power through an analysis and prediction module, formulates corresponding control strategies using a strategy formulation module, and based on the prediction results, performs unified intelligent coordinated control of photovoltaic power generation, battery energy storage, load power consumption and power grid according to the control strategy, uses a fault diagnosis module to monitor the power generation of the photovoltaic array, the charge state of the energy storage system and the power of the power load in real time, and determines the fault type and fault probability based on the neural network model of fault judgment, and performs fault warning and diagnosis.

[0024] The details are as follows: See also Figure 1 As shown, the present invention provides a highway distributed photovoltaic self-consistent energy supply system, including an analysis and prediction module 100, a strategy formulation module 200, a fault diagnosis module 300 and an optimization scheduling module 400; The analysis and prediction module 100 collects photovoltaic power generation data, energy storage data and load data through big data collection technology, extracts features from the collected data, collects meteorological data such as solar radiation intensity and battery temperature, uses the physical model method to predict photovoltaic power generation, uses the battery model method to predict the remaining available capacity of the energy storage system and the available charging and discharging power, and uses the regression model based on traffic flow and time factors to predict short-term power load fluctuations, and transmits the prediction results to the strategy formulation module 200; The strategy formulation module 200 formulates the photovoltaic power generation control strategy according to the prediction results of the analysis and prediction module 100 and the maximum power point tracking rule, determines the energy storage system control strategy through the state of charge control rule, determines the power load control strategy using the key load protection rule, and optimizes the control strategy using the fuzzy algorithm; The fault diagnosis module 300 monitors the power generation of the photovoltaic array, the state of charge of the energy storage system and the power of the power load in real time, establishes a neural network model for fault determination based on the historical data collected by the analysis and prediction module 100 using big data collection technology, and trains the neural network model, and then performs fault warning and diagnosis based on the output fault category and fault probability; The optimization scheduling module 400 uses the network-building microgrid support function, the bilateral service area centralized control function and the mutual assistance function according to the fault type determined by the fault diagnosis module 300 to reasonably allocate the excess power of the nearby microgrid support points to meet the power demand of the site.

[0025] like Figure 2 As shown, the analysis and prediction module 100 includes a data acquisition unit 110 and a model prediction unit 120; The data acquisition unit 110 collects the power generation power, voltage, current, light intensity, and temperature of the photovoltaic system, the state of charge, charge and discharge current, and temperature of the energy storage system, the power, voltage, and frequency of the grid access, and the power demand and power factor of the power load, and performs feature extraction on the collected data; The model prediction unit 120 predicts photovoltaic power generation by using a physical model method, predicts the remaining available capacity of the energy storage system and the available charging and discharging power by using a battery model method, and predicts short-term power load fluctuations by using a regression model based on traffic flow and time factors; The data acquisition unit 110 obtains real-time light intensity data by installing high-precision light intensity sensors in the photovoltaic power station. These sensors can be installed near the photovoltaic array, and the data acquisition frequency can be set according to actual needs, such as collecting data every few minutes. At the same time, the historical light intensity data of the local weather station and the future short-term (several hours to several days) light intensity forecast data are obtained. The forecast data source can be the numerical weather forecast model of a professional meteorological agency; Temperature sensors are installed on or near the surface of the PV modules to measure the actual operating temperature of the modules. In addition, local temperature forecast data is collected because the module temperature is closely related to the ambient temperature and is affected by factors such as light intensity. For example, the temperature of PV modules is usually higher than the ambient temperature by a certain amount, and the specific relationship can be determined based on empirical formulas or actual tests.

[0026] In order to better predict the photovoltaic power generation through the physical model method, the model prediction unit 120 predicts the photovoltaic power generation through the physical model method, and the physical model formula is:

[0027] in, is the power generation, is the photoelectric conversion efficiency of the photocell, is the area of ​​the photovoltaic panel, is the solar radiation intensity, is the temperature coefficient, is the actual temperature of the photovoltaic panel, is the reference temperature.

[0028] First, it is necessary to obtain accurate meteorological data, including light intensity, temperature, wind speed, cloud coverage, etc. This data can be obtained by installing a meteorological station near the photovoltaic power station or obtaining forecast data from the local meteorological department. Then, combined with the technical parameters of the photovoltaic modules, the meteorological data is substituted into the physical model for calculation. At the same time, considering the long-term impact of factors such as aging and dust coverage of photovoltaic modules on power generation, the model parameters are corrected by regularly testing the performance of photovoltaic modules.

[0029] In order to better predict the remaining available capacity of the energy storage system and the available charge and discharge power through the battery model method, the model prediction unit 120 predicts the remaining available capacity of the energy storage system and the available charge and discharge power through the battery model method, and its differential equation formula is:

[0030] in, is the terminal voltage of the battery, is the open circuit voltage of the battery, is the internal resistance, is the polarization capacitance, For current.

[0031] First, the parameters of the battery model need to be determined. This can be done by performing charge and discharge experiments on the battery, measuring parameters such as the battery terminal voltage at different charge and discharge currents and times, and then determining the model parameters through parameter identification methods (such as the least squares method). During operation, the terminal voltage and current of the battery are measured in real time, substituted into the battery model equation, and the remaining available capacity of the battery is calculated based on the initial capacity and the amount of charge and discharge of the battery. The available charge and discharge power can be determined based on the current state of the battery (such as SOC, temperature, etc.) and the performance curve of the battery (provided by the battery manufacturer). Generally speaking, the battery can provide higher discharge power at high SOC and suitable temperature, and the discharge power will be limited at low SOC and high or low temperature environments.

[0032] The strategy formulation module 200 includes a rule determination unit 210 and an optimization algorithm unit 220; The rule determination unit 210 formulates a photovoltaic power generation control strategy according to the maximum power point tracking rule, determines the energy storage system control strategy through the state of charge control rule, and determines the power load control strategy using the key load protection rule; The optimization algorithm unit 220 takes the light intensity and temperature as input, and after fuzzification, adjusts the working point of the photovoltaic inverter according to the predefined fuzzy rules, takes the state of charge and the power load demand as input to formulate the energy storage system control strategy, and takes the difference between the local power generation power and the power load and the grid access status as input to set the grid interaction rules; A series of fault diagnosis rules are formulated based on the working principle and performance parameters of the equipment. For example, for a photovoltaic inverter, when the output power is lower than a certain percentage of the normal power generation (such as 80%), and the light intensity and temperature are within the normal range, it may be an internal fault of the inverter; for an energy storage system, when the temperature of the battery pack exceeds the set safety temperature limit, or the battery state of charge (SOC) drops abnormally quickly, it may indicate a battery management system (BMS) failure or a problem with the battery cell.

[0033] The maximum power point tracking rule is that the output power of the photovoltaic cell is related to the light intensity, temperature and load characteristics, and its power-voltage characteristic curve is single-peaked. The purpose of the MPPT strategy is to allow the photovoltaic inverter to adjust its operating point in real time so that the photovoltaic array always operates near the maximum power point to obtain the maximum power generation efficiency; The perturbation observation method is used to periodically perturb the output voltage of the photovoltaic array with a small amplitude to observe the direction of power change. If the power increases, continue to perturb in that direction; if the power decreases, change the perturbation direction; In some cases, such as when the grid access power is limited, the energy storage system is full, or the local power load is low, it is necessary to limit the photovoltaic power generation power. This can prevent the waste of electricity or avoid impact on the grid. When power limitation is required, the energy management system (EMS) sends a control instruction to the photovoltaic inverter to reduce the power generation power by adjusting the output current or voltage of the inverter. For example, a proportional-integral controller can be used to calculate the control signal based on the difference between the set power limit value and the actual power generation, and adjust the output of the inverter to stabilize the power generation power within the limit range.

[0034] The charging and discharging mode is determined according to the state of charge (SOC) of the energy storage system. When the SOC is lower than the set lower limit, the energy storage system is controlled to enter the charging mode; when the SOC is higher than the set upper limit, the energy storage system is controlled to enter the discharging mode; when the SOC is between the upper and lower limits, the charging and discharging are determined according to the energy balance requirements of the system (such as power load and photovoltaic power generation). For example, when power consumption is low and photovoltaic power generation is sufficient, if the SOC has not reached the upper limit, the energy storage system will be charged first; when power consumption is high and photovoltaic power generation is insufficient, if the SOC has not reached the lower limit, the energy storage system will be controlled to discharge; The charging and discharging power of the energy storage system is reasonably allocated according to the dynamic changes of the system's power load demand and photovoltaic power generation power. For example, when the power load suddenly increases and the photovoltaic power generation power cannot meet the demand, the EMS calculates the appropriate discharge power based on the current SOC of the energy storage system and the maximum discharge power that can be provided to supplement the power gap. Similarly, when the photovoltaic power generation power is in excess, a reasonable charging power is allocated according to the charging capacity of the energy storage system.

[0035] In order to set the anti-backflow rule, the rule determination unit 210 installs a high-precision power sensor at the grid access point to monitor the power flow in real time, and formulates the anti-backflow rule by setting a backflow power threshold and a time threshold; By installing power sensors at the grid access point, the power flow is monitored in real time. When a reverse flow trend is detected (i.e., electric energy flows from the local system to the grid), the EMS takes prompt action. For example, it adjusts the photovoltaic power generation power (through MPPT control or power limitation) and the charging and discharging status of the energy storage system, preferentially stores excess electric energy in the energy storage system, or adjusts the working mode of local power equipment (such as delaying the working time of some non-critical loads) to ensure that the reverse flow time does not exceed the set threshold (such as 2 seconds); Combined with the photovoltaic power generation forecast and the power load forecast, the system operation mode can be adjusted in advance to avoid reverse flow. For example, if it is predicted that the photovoltaic power generation will increase significantly in the next period of time, while the local power load will not change much, the energy storage system can be controlled to enter the charging state in advance, or the grid dispatch center can be communicated with to increase the allowed power on the grid.

[0036] In order to finally determine the control strategy, the optimization algorithm unit 220 uses a decision tree algorithm to finally formulate the optimized control strategy; A decision tree is a tree structure, where each internal node is a test on an attribute, each branch is a test output, and each leaf node is a category or a value. In the formulation of power system control strategies, system state attributes (such as whether photovoltaic power generation is sufficient, whether energy storage SOC is sufficient, etc.) are used as nodes, and control strategy branches are generated by judging these attributes; Photovoltaic power generation control: The root node can be "Is the photovoltaic power generation greater than the local power load?" If so, the branch can be "Is the energy storage system full?" If not, the control strategy of the leaf node can be "Charge the energy storage system at a certain power, and transmit the remaining power to the grid or use it for local non-critical loads"; Energy storage system control: The root node is "Is the energy storage SOC lower than the lower limit?" If yes, the branch is "Is the photovoltaic power sufficient for charging?" If yes, the leaf node strategy is "Charge at maximum power"; if no, the leaf node strategy is "Get power from the grid for charging or reduce the power of non-critical loads"; Grid interaction rule setting: The root node is "Is there a reverse flow trend?" If yes, the branch is "Can the energy storage system absorb excess power?" If yes, the leaf node strategy is "Control the energy storage system to absorb power and adjust the photovoltaic power generation power"; if no, the leaf node strategy is "Adjust the power of local power equipment or limit power supply to the grid."

[0037] like Figure 3 As shown, the fault diagnosis module 300 includes a data monitoring unit 310 and a fault determination unit 320; The data monitoring unit 310 collects data covering various system operating states, including data of normal operating states and known fault states, and monitors the power generation power of the photovoltaic array, the charge state of the energy storage system, and the power of the power load in real time; The fault determination unit 320 uses the fault category and fault probability of the neural network model for fault determination to perform fault warning and diagnosis; In order to determine the minimum loss function of the neural network model for fault determination, the fault determination unit (320) uses a multi-classification cross entropy loss function to measure the difference between the model prediction probability distribution and the true probability distribution when establishing the neural network model for fault determination, and trains the neural network model with the minimum loss function as the target, wherein the calculation formula of the multi-classification cross entropy loss function is:

[0038] in, is the probability distribution of the model output, is the probability distribution of the true label, is the number of categories.

[0039] Generally, the data set is divided into a training set according to a certain ratio (such as 70%-80%) for training the neural network; 10%-15% is divided into a validation set for adjusting the model's hyperparameters (such as learning rate, number of hidden layer neurons, etc.) during the training process; the remaining 10%-20% is used as a test set to evaluate the performance of the final model; The training set data is input into the neural network, the output is calculated through forward propagation, and then the gradient is calculated using the back propagation algorithm based on the difference between the output and the true label (loss function), and the weights and bias of the neural network are updated. This process is repeated until the loss function converges or the preset number of training rounds is reached. During the training process, the validation set is used to monitor the performance of the model to prevent overfitting. If the loss function on the validation set starts to rise or the accuracy decreases, it means that overfitting may have occurred, and measures can be taken such as adding regularization terms, stopping training early, etc.

[0040] In order to optimize the neural network model for fault determination, when the fault determination unit 320 establishes the neural network model for fault determination, the Adam optimization algorithm is used to combine the advantages of the momentum method and the adaptive learning rate to calculate the first-order moment estimation and the second-order moment estimation of the gradient, and then the learning rate and the update parameters are adjusted according to the two estimates; The core idea of ​​the Adam optimization algorithm is to maintain an adaptive learning rate for each parameter, and use the first-order and second-order moment information of the gradient to accelerate convergence and reduce oscillations during training. After each training sample (or small batch sample) is input and back-propagated, the Adam optimization algorithm is used to update the model parameters according to the above steps. After multiple training cycles, the model parameters will gradually converge to a better value, thereby improving the model's ability to determine the fault type.

[0041] In summary, the working principle of this solution is as follows: The distributed photovoltaic self-consistent energy supply system of the highway predicts the photovoltaic power generation power, energy storage charge and load operating power through the analysis and prediction module 100, formulates the corresponding control strategy through the strategy formulation module 200, and performs unified intelligent coordinated control of photovoltaic power generation, battery energy storage, load power consumption and power grid according to the prediction results and the control strategy, so as to dynamically adjust the power supply ratio of battery energy storage, load power consumption and power grid according to the predicted photovoltaic power generation power, monitor the power generation power of the photovoltaic array, the charge state of the energy storage system and the power of the power load in real time through the fault diagnosis module 300, determine the fault type and fault probability according to the neural network model of fault judgment, perform fault warning and diagnosis, and reasonably allocate the excess electric energy of the nearby microgrid fulcrums by using the meshed microgrid fulcrum function, the centralized control function of the bilateral service area and the mutual assistance function, so as to meet the power demand of the site and realize the local consumption and complementation of electric energy.

[0042] The above shows and describes 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 by the above embodiments. The above embodiments and descriptions are only preferred examples of the present invention and are not intended to limit the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A highway distributed photovoltaic self-consistent energy supply system, characterized by: It includes an analysis and prediction module (100), a strategy formulation module (200), a fault diagnosis module (300) and an optimization scheduling module (400); The analysis and prediction module (100) collects photovoltaic power generation data, energy storage data and load data through big data collection technology, extracts features from the collected data, collects meteorological data such as solar radiation intensity and battery temperature, uses a physical model method to predict photovoltaic power generation, uses a battery model method to predict the remaining available capacity of the energy storage system and the available charging and discharging power, and uses a regression model based on traffic flow and time factors to predict short-term power load fluctuations, and transmits the prediction results to the strategy formulation module (200); The strategy formulation module (200) formulates a photovoltaic power generation control strategy according to the prediction result of the analysis and prediction module (100) and the maximum power point tracking rule, determines the energy storage system control strategy through the charge state control rule, determines the power load control strategy through the key load protection rule, and optimizes the control strategy through the fuzzy algorithm; The fault diagnosis module (300) monitors the power generation of the photovoltaic array, the charge state of the energy storage system and the power of the electrical load in real time, establishes a neural network model for fault determination based on the historical data collected by the analysis and prediction module (100) using big data collection technology, trains the neural network model, and then performs fault warning and diagnosis based on the output fault category and fault probability; The optimization scheduling module (400) rationally allocates excess electric energy of nearby microgrid branches based on the fault type determined by the fault diagnosis module (300), using the networking type microgrid support function, the double-side service area centralized control function, and the mutual assistance function, to meet the electric energy demand of the site.

2. The highway distributed photovoltaic self-consistent energy supply system according to claim 1 is characterized by: The analysis and prediction module (100) comprises a data acquisition unit (110) and a model prediction unit (120); The data collection unit (110) collects the power generation power, voltage, current, light intensity, and temperature of the photovoltaic system, the state of charge, charge and discharge current, and temperature of the energy storage system, the power, voltage, and frequency of the grid access, and the power demand and power factor of the power load, and performs feature extraction on the collected data; The model prediction unit (120) predicts photovoltaic power generation using a physical model method, predicts remaining available capacity and available charging and discharging power of an energy storage system using a battery model method, and predicts short-term power load fluctuations using a regression model based on traffic flow and time factors.

3. The highway distributed photovoltaic self-consistent energy supply system according to claim 2 is characterized by: The model prediction unit (120) predicts photovoltaic power generation through a physical model method, and the physical model formula is:

4. Among them, is the power generation, is the photoelectric conversion efficiency of the photocell, is the area of ​​the photovoltaic panel, is the solar radiation intensity, is the temperature coefficient, is the actual temperature of the photovoltaic panel, is the reference temperature.

5. The highway distributed photovoltaic self-consistent energy supply system according to claim 2 is characterized by: The model prediction unit (120) predicts the remaining available capacity and the available charging and discharging power of the energy storage system by using a battery model method, and the differential equation formula is:

6. Among them, is the terminal voltage of the battery, is the open circuit voltage of the battery, is the internal resistance, is the polarization capacitance, For current.

7. The highway distributed photovoltaic self-consistent energy supply system according to claim 1 is characterized by: The strategy formulation module (200) includes a rule determination unit (210) and an optimization algorithm unit (220); The rule determination unit (210) formulates a photovoltaic power generation control strategy according to a maximum power point tracking rule, determines an energy storage system control strategy through a state of charge control rule, and determines an electric load control strategy using a key load protection rule; The optimization algorithm unit (220) uses light intensity and temperature as inputs, and after fuzzification, adjusts the working point of the photovoltaic inverter according to predefined fuzzy rules, uses the state of charge and power load demand as inputs to formulate a control strategy for the energy storage system, and uses the difference between local power generation power and power load and the power grid access status as inputs to set power grid interaction rules.

8. The highway distributed photovoltaic self-consistent energy supply system according to claim 5 is characterized by: The rule determination unit (210) installs a high-precision power sensor at a grid access point to monitor the power flow in real time, and formulates a reverse flow prevention rule by setting a reverse flow power threshold and a time threshold.

9. The highway distributed photovoltaic self-consistent energy supply system according to claim 5 is characterized by: The optimization algorithm unit (220) uses a decision tree algorithm to finally formulate the optimized control strategy.

10. The highway distributed photovoltaic self-consistent energy supply system according to claim 1 is characterized by: The fault diagnosis module (300) comprises a data monitoring unit (310) and a fault determination unit (320); The data monitoring unit (310) collects data covering various system operating states, including data of normal operating states and known fault states, and monitors the power generation of the photovoltaic array, the charge state of the energy storage system, and the power of the electrical load in real time; The fault determination unit (320) uses the fault category and fault probability of the neural network model for fault determination to perform fault early warning and diagnosis.

11. The highway distributed photovoltaic self-consistent energy supply system according to claim 8, characterized in that: When the fault determination unit (320) establishes a neural network model for fault determination, a multi-classification cross entropy loss function is used to measure the difference between the model prediction probability distribution and the true probability distribution, and the neural network model is trained with the minimum loss function as the target, wherein the calculation formula of the multi-classification cross entropy loss function is:

12. Among them, is the probability distribution of the model output, is the probability distribution of the true label, is the number of categories.

13. The highway distributed photovoltaic self-consistent energy supply system according to claim 9, characterized in that: When the fault judgment unit (320) establishes a neural network model for fault judgment, it uses the Adam optimization algorithm to combine the advantages of the momentum method and the adaptive learning rate to calculate the first-order moment estimation and the second-order moment estimation of the gradient, and then adjusts the learning rate and updates the parameters according to the two estimates.

Images