Grid-connected and off-grid switching control system of grid-forming type energy storage converter

Abstract

The invention discloses a grid-connected and off-grid switching control system for a grid-forming type energy storage converter, and the system comprises a grid-connected operation state obtaining module which obtains the grid-connected operation state of a current power grid; the operation vector construction module is used for obtaining a power grid operation vector at the current moment; the label output module is used for outputting a power grid condition label in future rated time; the first execution module is used for executing the first switching instruction and switching the converter from a grid-connected mode to an off-grid mode; the continuous acquisition module is used for continuously acquiring the grid-connected operation state of the current power grid, and the second execution module is used for executing a second switching instruction and switching the converter from the off-grid mode to the grid-connected mode; according to the method, the pre-trained power grid prediction model is used for predicting the future power grid condition, the converter can realize grid-connected and off-grid switching of the constructed grid type energy storage converter based on the power grid parameters collected in real time and the constructed grid-connected operation state, and faults caused by power grid abnormity are avoided.

Description

Technical Field

[0001] The invention relates to an on-grid and off-grid switching control system, in particular to an on-grid and off-grid switching control system for a grid-connected energy storage converter. Background Art

[0002] With the rapid development of renewable energy and the widespread application of distributed energy systems, the role of energy storage systems in power systems is becoming increasingly important. As a key interface connecting energy storage devices and the power grid, the Grid-Forming Energy Storage Converter (GFESC) not only needs to have efficient energy conversion capabilities, but also must be able to smoothly switch between grid-connected and off-grid modes to ensure system stability and power quality. Traditional energy storage converters usually rely on fixed control strategies for mode switching. This static method is difficult to adapt to the complex and changing power grid environment, and is prone to instability during the switching process, affecting the reliability and safety of the system.

[0003] Patent document with patent publication number CN118589617A discloses a grid-connected energy storage inverter on-grid and off-grid switching control method and grid-connected energy storage inverter, which can realize the switching of the grid-connected energy storage inverter from the grid-connected zero discharge state to the off-grid full power discharge state, and ensure the output voltage is stable and has good dynamic characteristics. However, the prior art including the above patent solution still relies on fixed control strategies for mode switching, uses preset fixed parameters for switching between grid-connected and off-grid modes, lacks the ability to dynamically respond to real-time grid conditions, and cannot effectively deal with sudden grid anomalies. Summary of the invention

[0004] In view of the deficiencies in the prior art, the present invention provides a grid-connected energy storage inverter on-grid and off-grid switching control system, which solves the technical problems raised in the background technology by real-time collection of key parameters such as grid voltage, AC frequency and phase angle.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solutions:

[0006] A grid-connected and off-grid switching control system for a grid-connected energy storage converter, comprising:

[0007] The grid-connected operation status acquisition module is used to collect the voltage, AC frequency and phase angle of the power grid in real time to obtain the current grid-connected operation status;

[0008] An operation vector construction module is used to perform feature engineering on the grid-connected operation state of the current power grid to obtain a power grid operation vector at the current moment;

[0009] A label output module, used to input the power grid operation vector at the current moment into a pre-trained power grid prediction model, and output a label of the power grid status within a future rated time; if the output power grid status label within the future rated time is abnormal, generate a first switching instruction;

[0010] A first execution module, configured to execute a first switching instruction to switch the converter from a grid-connected mode to an off-grid mode;

[0011] A continuous acquisition module is used to continuously acquire the grid-connected operation status of the current power grid and output a grid status tag within a future rated time; if the grid status tag within the future rated time is healthy, a second switching instruction is generated;

[0012] The second execution module is used to execute a second switching instruction to switch the converter from an off-grid mode to a grid-connected mode.

[0013] In some embodiments, executing the first switching instruction to switch the converter from the grid-connected mode to the off-grid mode includes:

[0014] S5-1, executing a first parameter adjustment instruction to gradually adjust the output parameters of the converter so that the converter transitions to an off-grid mode;

[0015] S5-2, executing the first mode switching instruction, cutting off the connection between the converter and the main power grid, and synchronously starting the off-grid mode of the converter;

[0016] S5-3. When the converter is in off-grid mode, dynamically adjust the output parameters of the converter to maintain the stability of the output parameters of the local microgrid.

[0017] In some embodiments, executing the first parameter adjustment instruction to gradually adjust the output parameter of the converter so as to transition the converter to an off-grid mode includes:

[0018] S5-1-1, gradually adjust the output voltage of the converter to make it gradually approach the preset voltage of the off-grid mode;

[0019] S5-1-2, gradually adjust the output AC power frequency of the converter to make it gradually approach the preset AC power frequency of the off-grid mode;

[0020] S5-1-3. Gradually adjust the output phase angle of the converter so that the waveform of its output current or output voltage is consistent with the preset waveform in the off-grid mode.

[0021] In some embodiments, when the converter is in an off-grid mode, dynamically adjusting the output parameters of the converter to maintain the stability of the output parameters of the local microgrid includes:

[0022] S5-3-1, dynamically control the output voltage of the converter according to the preset voltage in off-grid mode,

[0023] S5-3-2, dynamically controlling the output AC power frequency of the converter according to the preset AC power frequency in the off-grid mode;

[0024] S5-3-3, dynamically controlling the output phase angle of the converter according to the preset waveform in the off-grid mode;

[0025] S5-3-4. According to a preset control strategy in off-grid mode, dynamically control the in-phase current component output by the converter to the local load; the preset control strategy is characterized by: ensuring that the total output current waveform of the converter in off-grid mode remains in phase with the preset voltage waveform;

[0026] S5-3-5. Dynamically control the in-phase current component output by each energy storage unit to the local load according to the demand of the local load.

[0027] In some embodiments, dynamically controlling the in-phase current component output by each energy storage unit to the local load according to the demand of the local load includes:

[0028] S5-3-5-1, the local load generates a real-time power request according to the real-time power demand;

[0029] S5-3-5-2, in response to the real-time power request, generating a dynamic power adjustment instruction;

[0030] S5-3-5-3. Execute dynamic power adjustment instructions to control the in-phase current component output by each energy storage unit to the local load.

[0031] In some embodiments, the grid-connected operation status of the current power grid is continuously acquired, and a grid status tag within a future rated time is output; if the grid status tag within the future rated time is healthy, a second switching instruction is generated, including:

[0032] S6-1, continuously monitor the voltage, AC frequency and phase angle of the power grid to determine the off-grid operation status of the current power grid;

[0033] S6-2, evaluating whether the grid connection conditions are met according to the off-grid operation status of the current grid;

[0034] S6-3, if the grid connection condition is met, execute the second switching instruction;

[0035] S6-4, executing the second switching instruction to switch the converter from the off-grid mode to the grid-connected mode.

[0036] In some embodiments, executing the second switching instruction to switch the converter from an off-grid mode to a grid-connected mode includes:

[0037] S7-1, gradually adjust the output voltage of the converter to make it gradually approach the current grid voltage;

[0038] S7-2, gradually adjust the output AC power frequency of the converter to make it gradually approach the current AC power frequency of the power grid;

[0039] S7-3. Gradually adjust the output phase angle of the converter so that the waveform of its output current is in phase with the voltage waveform of the current power grid.

[0040] In some embodiments, the modeling step of the power grid prediction model includes:

[0041] S4-1, collecting the grid voltage, AC frequency and phase angle at historical time points;

[0042] S4-2, according to the grid voltage, AC frequency and phase angle at the historical time point, define a grid status label binary label at the historical time point, including an abnormal label and a healthy label;

[0043] S4-3, performing feature processing on the grid voltage, AC power frequency and phase angle at the historical time point to obtain a grid operation vector at the historical time point;

[0044] S4-4, using the power grid operation vector at the historical time point as an input feature and the power grid status label binary classification label as a true label to construct a training sample at the historical time point;

[0045] S4-5. Define a time window of fixed length, and collect training samples at multiple historical time points within the time window; slide the window several times along the time axis, and re-collect training samples within the corresponding time period after each slide to form a training set;

[0046] S4-6, extracting training samples from the training set in batches, inputting them into the time series model, and outputting prediction labels;

[0047] S4-7. Calculate the probability distribution of the predicted label and the cross entropy loss of the true label;

[0048] Among them, the cross entropy loss function is:

[0049]

[0050] L represents the sum of the cross entropy losses of all training samples in the training set, N represents the total number of training samples, and y i represents the true label of the i-th training sample, Represents the predicted probability of the i-th training sample output by the model;

[0051] S4-8, iteratively updating the model parameters of the timing model until the cross entropy loss is minimized to obtain the power grid prediction model;

[0052] When the training is completed, the obtained model parameters are the optimal parameters after minimizing the cross entropy loss, and the model with the optimal parameters is exported as the final power grid prediction model.

[0053] In some of the embodiments, a time window of fixed length is defined, and training samples of multiple historical time points are collected within the time window; and the window is slid along the time axis for several times, and training samples within the corresponding time period are recollected after each sliding, and combined into a training set, including:

[0054] S4-5-1. Set a time window of fixed length T;

[0055] S4-5-2, collecting training samples at multiple historical time points at the initial position of the time window;

[0056] S4-5-3, sliding the time window Δt forward along the time axis to obtain a time window including the first sliding position;

[0057] S4-5-4, re-collecting training samples at multiple historical time points within the time window at the first sliding position;

[0058] S4-5-5, repeat steps S4-5-3 and S4-5-4 until the preset historical duration is covered; S4-5-6, combine the training samples within the preset historical duration into the training set.

[0059] The present invention provides a grid-connected energy storage converter on-grid and off-grid switching control system, which has the following beneficial effects:

[0060] The pre-trained grid prediction model of the present invention is used to predict the future status of the grid. The converter can realize the on-grid and off-grid switching of the grid-connected energy storage converter based on the real-time collected grid parameters and the constructed grid-connected operation status, thereby avoiding failures caused by grid abnormalities.

[0061] On the other hand, when the energy storage inverter is in off-grid mode, the system can dynamically adjust the output parameters according to the needs of the local load, ensuring maximum power transmission efficiency and minimizing reactive power, thereby improving energy utilization efficiency.

[0062] In addition, the present invention effectively captures the dynamic changes and trends of time series data through the modeling process of the time series model for predicting the health status of the power grid, combined with the sliding window strategy, so that the model can learn richer patterns and laws. Continuously sliding the time window and re-collecting training samples ensures that the data of each time period is fully considered, avoiding the limitations of data at a single time point, and improving the generalization of the pre-trained power grid prediction model. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 This is a structural block diagram of a grid-connected energy storage converter on-grid and off-grid switching control system of the present invention;

[0064] Figure 2 It is a schematic diagram of the execution of the grid-connected and off-grid switching control system of the grid-connected energy storage converter of the present invention;

[0065] Figure 3 A schematic diagram of the modeling steps of the power grid prediction model of the present invention;

[0066] Figure 4 Schematic diagram of the steps for collecting the training set described in the present invention. DETAILED DESCRIPTION

[0067] The following will be combined with the drawings in the embodiments of 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.

[0068] Example 1: Please refer to Figure 1 - Figure 2 The present invention provides a grid-connected energy storage converter on-grid and off-grid switching control system, comprising:

[0069] The grid-connected operation status acquisition module is used to collect the voltage, AC frequency and phase angle of the power grid in real time to obtain the current grid-connected operation status of the power grid; the current grid-connected operation status of the power grid is obtained through sensors installed at key nodes of the power grid.

[0070] A parameter initialization module, used for calibrating the output parameters of the converter according to the grid-connected operation state of the current power grid;

[0071] Specifically, the converter can use a PID controller to calibrate the output voltage, frequency and phase angle to synchronize it with the grid. The output voltage is adjusted by adjusting the output amplitude of the inverter, the output frequency is synchronized with the grid frequency through a phase-locked loop (PLL), and the output phase angle is adjusted by a phase detector to adjust the current waveform.

[0072] Exemplarily, the output parameters include:

[0073] Output voltage, used to adjust the voltage amplitude of the converter output to make it consistent with the grid voltage;

[0074] Output AC power frequency, used to adjust the AC power frequency output by the converter to make it consistent with the AC power frequency of the power grid;

[0075] Output phase angle, used to adjust the phase angle of the converter output current or voltage waveform to ensure that it remains the same as the grid voltage waveform;

[0076] Specifically, according to the current grid-connected operation status, the converter will synchronize with the grid by adjusting its output voltage, frequency and phase angle. The converter first adjusts its output voltage according to the grid voltage amplitude to ensure that the two are consistent. Next, the converter will adjust its output frequency to make it consistent with the AC frequency of the grid, thereby avoiding system instability caused by frequency differences. Finally, the output phase angle is adjusted to ensure that the current or waveform output by the converter is synchronized with the grid voltage waveform.

[0077] An operation vector construction module is used to perform feature engineering on the grid-connected operation state of the current power grid to obtain a power grid operation vector at the current moment;

[0078] In this embodiment, the voltage, AC frequency, and phase angle of the power grid are raw data. Feature engineering converts the current grid-connected operation status (such as voltage, frequency, and phase angle) into a formalized grid operation vector that is convenient for subsequent analysis and decision-making. Feature engineering includes preprocessing steps such as standardization and normalization to ensure that the data meets the relevant input requirements.

[0079] A label output module, used to input the current power grid operation vector into a pre-trained power grid prediction model, and output a label of the power grid status within a future rated time;

[0080] The model is trained based on historical data and can predict the grid status labels within the future rated time. The prediction results are used to guide the converter to make corresponding adjustment preparations in advance to avoid failures caused by grid abnormalities. By predicting the grid status labels of the future grid, the converter can make corresponding adjustment preparations in advance to avoid failures caused by grid abnormalities. If the grid status labels within the future rated time are predicted to be abnormal, the converter can respond quickly and take measures to switch to off-grid mode to ensure the stability of the system.

[0081] If the outputted power grid status label within the future rated time is abnormal, a first switching instruction is generated;

[0082] Exemplarily, the first switching instruction includes:

[0083] The first parameter adjustment instruction is used to gradually adjust the output parameters of the converter (including voltage, AC frequency, and phase angle) to make them close to the output parameters of the converter in the off-grid mode;

[0084] The first mode adjustment instruction is used to switch the converter from the grid-connected mode to the off-grid mode.

[0085] The first execution module is used to execute a first switching instruction to switch the converter from a grid-connected mode to an off-grid mode.

[0086] A continuous acquisition module is used to continuously acquire the grid-connected operation status of the current power grid and output a grid status tag within a future rated time; if the grid status tag within the future rated time is healthy, a second switching instruction is generated;

[0087] The second execution module is used to execute a second switching instruction to switch the converter from an off-grid mode to a grid-connected mode.

[0088] Exemplarily, the second switching instruction includes:

[0089] The second parameter adjustment instruction is used to gradually adjust the output parameters of the converter in the off-grid mode to make them close to the output parameters of the converter in the grid-connected mode;

[0090] The second mode adjustment instruction is used to switch the converter from the off-grid mode to the grid-connected mode.

[0091] The specific execution steps of the first execution module include:

[0092] S5-1, executing a first parameter adjustment instruction to gradually adjust the output parameters of the converter so that the converter transitions to an off-grid mode;

[0093] S5-2, executing the first mode switching instruction, cutting off the connection between the converter and the main power grid, and synchronously starting the off-grid mode of the converter;

[0094] S5-3, when the converter is in off-grid mode, dynamically adjusting the output parameters of the converter to maintain the stability of the output parameters of the local microgrid;

[0095] Specifically, the present invention provides a grid-connected energy storage converter on-grid and off-grid switching control system, including real-time acquisition of grid parameters, calibration of converter output parameters, feature engineering, prediction of future grid status labels, generation of switching instructions, execution of parameter adjustment and mode switching steps. Through these steps, it is ensured that the converter can switch smoothly between on-grid and off-grid modes, maintaining the stability of the system and the quality of power.

[0096] Furthermore, the step S5-1 specifically includes:

[0097] S5-1-1, gradually adjust the output voltage of the converter to make it gradually approach the preset voltage of the off-grid mode;

[0098] The step S5-1-1 gradually matches the output voltage of the converter with the preset voltage in the off-grid mode. When connected to the grid, the output voltage of the converter is usually synchronized with the grid voltage, while in the off-grid mode, the converter needs to generate a stable output voltage by itself to ensure that it is consistent with the load demand.

[0099] S5-1-2, gradually adjust the output AC power frequency of the converter to make it gradually approach the preset AC power frequency of the off-grid mode;

[0100] In step S5-1-2, the output AC frequency is determined by the grid in the grid-connected mode, and in the off-grid mode, the converter needs to output current according to the preset AC frequency, which is usually maintained by the power provided by the energy storage unit. Therefore, gradually adjusting the AC frequency is the key to ensure the stable operation of the converter.

[0101] S5-1-3. Gradually adjust the output phase angle of the converter so that the waveform of its output current or output voltage is consistent with the preset waveform in the off-grid mode.

[0102] Furthermore, the step S5-3 specifically includes:

[0103] S5-3-1, dynamically control the output voltage of the converter according to the preset voltage in off-grid mode,

[0104] S5-3-2, dynamically controlling the output AC power frequency of the converter according to the preset AC power frequency in the off-grid mode;

[0105] S5-3-3, dynamically controlling the output phase angle of the converter according to the preset waveform in the off-grid mode;

[0106] S5-3-4. According to a preset control strategy in off-grid mode, dynamically control the in-phase current component output by the converter to the local load; the preset control strategy is characterized by: ensuring that the total output current waveform of the converter in off-grid mode remains in phase with the preset voltage waveform;

[0107] Specifically: In grid-connected mode, the goal of the converter is to keep its output current in phase with the grid voltage waveform to ensure maximum power transmission efficiency. Specifically, when the total output current of the converter is in phase with the grid voltage waveform, the phase difference between the current and voltage is 0 degrees, and the power factor is close to 1, indicating that most of the electrical energy is effectively transmitted.

[0108] The relationship between active power and reactive power can be expressed by the grid power expression:

[0109] P1 = VIcos(θ);

[0110] Wherein, P1 is the active power in grid-connected mode, V is the grid voltage, I is the total output current of the converter, and θ is the phase difference between the grid voltage and the total output current of the converter. In grid-connected mode, the goal of the converter is to make the total output current of the converter in phase with the grid voltage, that is, the phase difference θ = 0, so that the power factor is close to 1, maximizing the active power transmission while minimizing the reactive power.

[0111] In off-grid mode, the preset control strategy means that the converter is no longer synchronized with the grid voltage, but generates its own preset voltage waveform. At this time, the control strategy is to ensure that the total output current of the converter is in phase with the preset voltage waveform generated by the converter itself. At this time, the relationship between active power and reactive power can be expressed by the off-grid power expression;

[0112] The off-grid power expression is:

[0113] P2=V inverter I inverter cos(θ);

[0114] Where P2 is the active power in off-grid mode, V inverter is the voltage output by the inverter (i.e. the preset voltage in off-grid mode), I inverter is the total output current of the converter in off-grid mode;

[0115] When θ = 0, the active power P 2 in off-grid mode is maximized.

[0116] S5-3-5, dynamically controlling the in-phase current component output by each energy storage unit to the local load according to the demand of the local load;

[0117] Specifically, local loads refer to electrical devices or systems directly connected to GFESC (grid-type energy storage converter) or local microgrids. These loads can be any equipment that consumes electrical energy, such as lighting systems, motors, household appliances, etc.; local load demand refers to the amount and characteristics of electrical energy required by the above electrical equipment at a certain moment (such as power, current, voltage, etc.).

[0118] Furthermore, the specific steps of step S5-3-5 also include:

[0119] S5-3-5-1, the local load generates a real-time power request according to the real-time power demand;

[0120] Local loads (such as lighting systems, motors, household appliances, etc.) will generate a real-time power request based on their current power demand. It reflects the power characteristics (such as power, current, voltage, etc.) required by the local load at a certain moment. For example, when a motor needs to start, it may issue a higher power request; when the lighting system is in a low brightness state, it may only require a lower power.

[0121] S5-3-5-2, in response to the real-time power request, generating a dynamic power adjustment instruction;

[0122] After receiving the real-time power request, the converter generates the corresponding dynamic power adjustment instruction according to the needs of the local load. The dynamic power adjustment instruction is used to instruct the energy storage unit on how to adjust its output parameters to meet the needs of the load. For example, if a local load requests to increase power, the converter will generate an instruction to require the energy storage unit to increase the output power; conversely, if the load requests to reduce power, it will generate an instruction to reduce the output power.

[0123] S5-3-5-3. Execute dynamic power adjustment instructions to control the in-phase current component output by each energy storage unit to the local load.

[0124] The converter controls the in-phase current component of each energy storage unit output to the local load according to the generated dynamic power adjustment command. This ensures that the total output current waveform of the converter is in phase with the preset voltage waveform, thereby maximizing active power transmission while minimizing reactive power, ensuring efficient operation of the system and power quality. For example, when a certain energy storage unit is required to increase its output power, its output current will be adjusted accordingly to ensure that the current waveform is synchronized with the preset voltage waveform to achieve the best power factor and energy transmission efficiency.

[0125] Specifically, step S8-5 describes how the converter performs dynamic power adjustment according to the real-time power demand of the local load in the off-grid mode, including generating a real-time power request, generating a dynamic power adjustment instruction, and executing these instructions to control the in-phase current component output by the energy storage unit to the local load.

[0126] The steps of the continuous acquisition module are similar to those of the grid-connected operation status acquisition module, and both include:

[0127] S6-1, continuously monitor the voltage, AC frequency and phase angle of the power grid to determine the off-grid operation status of the current power grid;

[0128] In off-grid mode, the converter continuously monitors the voltage, AC frequency and phase angle of the grid to determine whether the current grid is still in off-grid operation.

[0129] However, in the step of the continuous acquisition module, compared with the step of the grid-connected operation status acquisition module, step S6-2 is added; specifically,

[0130] S6-2, evaluating whether the grid connection conditions are met according to the off-grid operation status of the current grid;

[0131] The grid connection conditions include whether the grid voltage, AC frequency and phase angle are within the preset range, and whether the grid is stable. Only when all conditions are met will the next step of parameter adjustment and mode switching be considered.

[0132] S6-3, if the grid connection conditions are met, execute the second switching instruction; gradually adjust the output parameters of the converter to synchronize them with the voltage, AC frequency and phase angle of the current grid to ensure a smooth transition;

[0133] S6-4, executing the second switching instruction to switch the converter from the off-grid mode to the grid-connected mode.

[0134] In this embodiment, step S7 further describes the specific steps of switching from the off-grid mode to the grid-connected mode, including continuously monitoring grid parameters, evaluating grid-connected conditions, gradually adjusting converter output parameters to achieve synchronization, and finally executing the first mode switching instruction.

[0135] Furthermore, the specific execution steps of the second execution module include:

[0136] S7-1, gradually adjust the output voltage of the converter to make it gradually close to the current grid voltage; so that the output voltage of the converter gradually matches the current grid voltage, ensuring the voltage amplitude is consistent when connected to the grid

[0137] S7-2. Gradually adjust the output AC power frequency of the converter to make it gradually close to the current grid AC power frequency. Similar to the output voltage, the output AC power frequency must be synchronized with the grid AC power frequency in the grid-connected mode. Gradually adjusting the AC power frequency is the key to ensuring the stable operation of the converter.

[0138] S7-3, gradually adjusting the output phase angle of the converter so that the waveform of its output current is in phase with the voltage waveform of the current power grid;

[0139] The phase angle is adjusted to ensure that the current waveform of the converter is consistent with the voltage waveform of the current grid. This is to ensure that in grid-connected mode, the converter can be correctly synchronized with the grid and the current and voltage remain synchronized.

[0140] In this embodiment, the pre-trained grid prediction model of the present invention is used to predict the future grid conditions. The converter can realize the grid-connected and off-grid switching of the grid-connected energy storage converter based on the real-time collected grid parameters and the grid-connected operation status constructed by it, thereby avoiding failures caused by grid abnormalities. In off-grid mode, the system can dynamically adjust the output parameters according to the needs of the local load, ensure maximum power transmission efficiency and minimize reactive power, and improve energy utilization efficiency.

[0141] Example 2, see Figure 1 - to 4, the technical solution of this embodiment 2 is different from that of embodiment 1 in that the modeling steps of the power grid prediction model are disclosed, including:

[0142] S4-1, collecting the grid voltage, AC frequency and phase angle at historical time points;

[0143] S4-2, according to the grid voltage, AC frequency and phase angle at the historical time point, define a grid status label binary label at the historical time point, including an abnormal label and a healthy label;

[0144] S4-3, performing feature processing on the grid voltage, AC power frequency and phase angle at the historical time point to obtain a grid operation vector at the historical time point;

[0145] S4-4, using the power grid operation vector at the historical time point as an input feature and the power grid status label binary classification label as a true label to construct a training sample at the historical time point;

[0146] S4-5. Define a time window of fixed length, and collect training samples at multiple historical time points within the time window; slide the window several times along the time axis, and re-collect training samples within the corresponding time period after each slide to form a training set;

[0147] S4-6, extracting training samples from the training set in batches, inputting them into the time series model, and outputting prediction labels;

[0148] S4-7. Calculate the probability distribution of the predicted label and the cross entropy loss of the true label;

[0149] Among them, the cross entropy loss function is:

[0150]

[0151] L represents the sum of the cross entropy losses of all training samples in the training set, N represents the total number of training samples, and y i represents the true label of the i-th training sample, Represents the predicted probability of the i-th training sample output by the model;

[0152] S4-8, iteratively updating the model parameters of the timing model until the cross entropy loss is minimized to obtain the power grid prediction model.

[0153] Exemplarily, the specific sub-steps of step S4-8 may be performed with reference to the prior art. They may include:

[0154] Initialize model parameters (weights and biases); usually initialize randomly or use a preset method.

[0155] Forward propagation: input the training samples into the time series model and calculate the probability distribution of the output; and calculate the cross entropy loss based on the true label and the predicted probability.

[0156] Gradient calculation: Use the chain rule to calculate the gradient (partial derivative) of the cross entropy loss for each model parameter, which is the back-propagation process.

[0157] Parameter update: Update the model parameters according to the calculated gradients.

[0158] Repeat the above steps (forward propagation, gradient calculation, parameter update) until the cross entropy loss is minimized;

[0159] When the training is completed, the obtained model parameters are the optimal parameters after minimizing the cross entropy loss, and the model with the optimal parameters is exported as the final power grid prediction model.

[0160] Specifically, the present invention further describes the modeling steps of the power grid prediction model, including data collection, label definition, feature processing, training sample construction, sliding window strategy, model training and loss function calculation, to ensure that the power grid prediction model can accurately predict the grid status label of the power grid within the future rated time, so as to take measures in advance to ensure the stability and reliability of the system.

[0161] Wherein, the step S4-5 specifically includes:

[0162] S4-5-1. Set a time window of fixed length T; the fixed length T can be hourly or daily to ensure that the data in each time period has a similar time span a

[0163] S4-5-2. Collect training samples of multiple historical time points at the initial position of the time window; the initial position represents the earliest time point, and the training samples include characteristic data such as grid voltage, AC frequency and phase angle and their corresponding grid status labels.

[0164] S4-5-3, slide the time window Δt forward along the time axis to obtain a time window including the first sliding position; this ensures that the time window after each sliding still covers a time period of the same length, but includes different historical time points.

[0165] S4-5-4, re-collecting training samples of multiple historical time points within the time window at the first sliding position; this ensures that new and independent training samples can be obtained after each sliding, thereby increasing the diversity and representativeness of the training samples.

[0166] S4-5-5. Repeat steps S4-5-3 and S4-5-4 until the preset historical duration is covered; for example, if the preset historical duration is one month, it is necessary to continuously slide the time window and collect samples until all time periods within this month are covered.

[0167] S4-5-6. Combine the training samples within a preset historical time period into the training set.

[0168] Specifically, step S4-5 further describes the collection and combination of training samples through the sliding window strategy in the process of building the power grid prediction model. Specifically, it includes setting a time window of fixed length, collecting training samples at the initial position of the time window, sliding the time window forward along the time axis, re-collecting training samples, and finally combining all training samples within the preset historical time length into a complete training set. This ensures the diversity of training samples and improves the generalization ability of the model.

[0169] In this embodiment, the present invention ensures that the power grid prediction model can accurately predict the status label of the power grid within the future rated time through the modeling process of the time series model that predicts the health status of the power grid, so as to take measures in advance to ensure the stability and reliability of the system. The sliding window strategy ensures that the training samples are not only sufficient in number, but also diverse and representative, thereby improving the generalization ability of the model. The sliding window strategy can effectively capture the dynamic changes and trends of time series data, so that the model can learn richer patterns and laws. Continuously sliding the time window and re-collecting training samples ensures that the data of each time period is fully considered, avoiding the limitations of data at a single time point.

[0170] The above embodiments can be implemented in whole or in part by software, hardware, firmware or other arbitrary combinations. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium, for example, the computer instructions can be transmitted from a website site, computer, server or data center to another website site, computer, server or data center by wired (e.g., infrared, wireless, microwave, etc.) mode.

[0171] The computer-readable storage medium may be any available medium that can be accessed by a computer or a data storage device such as a server or a data center that includes one or more available media. The available medium may be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium. The semiconductor medium may be a solid-state hard disk.

[0172] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division of a waterway underwater terrain change analysis system and method. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0173] The above description is only a specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application.

Claims

1. A grid-connected energy storage converter on-grid and off-grid switching control system, characterized in that: include: The grid-connected operation status acquisition module is used to collect the voltage, AC frequency and phase angle of the power grid in real time to obtain the current grid-connected operation status; An operation vector construction module is used to perform feature engineering on the grid-connected operation state of the current power grid to obtain a power grid operation vector at the current moment; A label output module, used to input the power grid operation vector at the current moment into a pre-trained power grid prediction model, and output a label of the power grid status within a future rated time; if the output power grid status label within the future rated time is abnormal, generate a first switching instruction; A first execution module, configured to execute a first switching instruction to switch the converter from a grid-connected mode to an off-grid mode; A continuous acquisition module is used to continuously acquire the grid-connected operation status of the current power grid and output the grid status label within a future rated time; If the grid status label within the future rated time is healthy, a second switching instruction is generated; The second execution module is used to execute a second switching instruction to switch the converter from an off-grid mode to a grid-connected mode.

2. A grid-connected energy storage converter on-grid and off-grid switching control system according to claim 1, characterized in that: Executing a first switching instruction to switch the converter from a grid-connected mode to an off-grid mode includes: S5-1, executing a first parameter adjustment instruction to gradually adjust the output parameters of the converter so that the converter transitions to an off-grid mode; S5-2, executing the first mode switching instruction, cutting off the connection between the converter and the main power grid, and synchronously starting the off-grid mode of the converter; S5-3. When the converter is in off-grid mode, dynamically adjust the output parameters of the converter to maintain the stability of the output parameters of the local microgrid.

3. A grid-connected energy storage converter on-grid and off-grid switching control system according to claim 1, characterized in that: Executing the first parameter adjustment instruction to gradually adjust the output parameter of the converter so as to transition the converter to an off-grid mode includes: S5-1-1, gradually adjust the output voltage of the converter to make it gradually approach the preset voltage of the off-grid mode; S5-1-2, gradually adjust the output AC power frequency of the converter to make it gradually approach the preset AC power frequency of the off-grid mode; S5-1-3. Gradually adjust the output phase angle of the converter so that the waveform of its output current or output voltage is consistent with the preset waveform in the off-grid mode.

4. A grid-connected energy storage converter on-grid and off-grid switching control system according to claim 1, characterized in that: When the converter is in an off-grid mode, dynamically adjusting the output parameters of the converter to maintain the stability of the output parameters of the local microgrid includes: S5-3-1, dynamically control the output voltage of the converter according to the preset voltage in off-grid mode, S5-3-2, dynamically controlling the output AC power frequency of the converter according to the preset AC power frequency in the off-grid mode; S5-3-3, dynamically controlling the output phase angle of the converter according to the preset waveform in the off-grid mode; S5-3-4. According to a preset control strategy in off-grid mode, dynamically control the in-phase current component output by the converter to the local load; the preset control strategy is characterized by: ensuring that the total output current waveform of the converter in off-grid mode remains in phase with the preset voltage waveform; S5-3-5. Dynamically control the in-phase current component output by each energy storage unit to the local load according to the demand of the local load.

5. A grid-connected energy storage converter on-grid and off-grid switching control system according to claim 1, characterized in that: According to the needs of the local load, dynamically control the in-phase current component output by each energy storage unit to the local load, including: S5-3-5-1, the local load generates a real-time power request according to the real-time power demand; S5-3-5-2, in response to the real-time power request, generating a dynamic power adjustment instruction; S5-3-5-3. Execute dynamic power adjustment instructions to control the in-phase current component output by each energy storage unit to the local load.

6. A grid-connected energy storage converter on-grid and off-grid switching control system according to claim 1, characterized in that: Continuously obtain the current grid-connected operation status and output the grid status label within the future rated time; If the grid status label within the future rated time is healthy, a second switching instruction is generated, including: S6-1, continuously monitor the voltage, AC frequency and phase angle of the power grid to determine the off-grid operation status of the current power grid; S6-2, evaluating whether the grid connection conditions are met according to the off-grid operation status of the current grid; S6-3: If the grid connection condition is met, execute the second switching instruction.

7. A grid-connected energy storage converter on-grid and off-grid switching control system according to claim 1, characterized in that: Executing a second switching instruction to switch the converter from an off-grid mode to a grid-connected mode includes: S7-1, gradually adjust the output voltage of the converter to make it gradually approach the current grid voltage; S7-2, gradually adjust the output AC power frequency of the converter to make it gradually approach the current AC power frequency of the power grid; S7-3. Gradually adjust the output phase angle of the converter so that the waveform of its output current is in phase with the voltage waveform of the current power grid.

8. A grid-connected energy storage converter on-grid and off-grid switching control system according to claim 1, characterized in that: The modeling steps of the power grid prediction model include: S4-1, collecting the grid voltage, AC frequency and phase angle at historical time points; S4-2, according to the grid voltage, AC frequency and phase angle at the historical time point, define a grid status label binary label at the historical time point, including an abnormal label and a healthy label; S4-3, performing feature processing on the grid voltage, AC power frequency and phase angle at the historical time point to obtain a grid operation vector at the historical time point; S4-4, using the power grid operation vector at the historical time point as an input feature and the power grid status label binary classification label as a true label to construct a training sample at the historical time point; S4-5. Define a time window of fixed length, and collect training samples at multiple historical time points within the time window; slide the window several times along the time axis, and re-collect training samples within the corresponding time period after each slide to form a training set; S4-6, extracting training samples from the training set in batches, inputting them into the time series model, and outputting prediction labels; S4-7. Calculate the probability distribution of the predicted label and the cross entropy loss of the true label; Among them, the cross entropy loss function is: L represents the sum of the cross entropy losses of all training samples in the training set, N represents the total number of training samples, and y i represents the true label of the i-th training sample, Represents the predicted probability of the i-th training sample output by the model; S4-8, iteratively updating the model parameters of the timing model until the cross entropy loss is minimized to obtain the power grid prediction model; When the training is completed, the obtained model parameters are the optimal parameters after minimizing the cross entropy loss, and the model with the optimal parameters is exported as the final power grid prediction model.

9. A grid-connected energy storage converter on-grid and off-grid switching control system according to claim 2, characterized in that: Define a time window of fixed length and collect training samples at multiple historical time points within the time window; The window is slid along the time axis several times, and the training samples in the corresponding time period are collected again after each slide, and combined into a training set, including: S4-5-1. Set a time window of fixed length T; S4-5-2, collecting training samples at multiple historical time points at the initial position of the time window; S4-5-3, sliding the time window Δt forward along the time axis to obtain a time window including the first sliding position; S4-5-4, re-collecting training samples at multiple historical time points within the time window at the first sliding position; S4-5-5, repeat steps S4-5-3 and S4-5-4 until the preset historical duration is covered; S4-5-6, combine the training samples within the preset historical duration into the training set.

Images