A simulation system for new energy power generation

Abstract

This invention relates to the field of power grid technology, specifically to a simulation system for new energy power generation, comprising: a state simulation module, a sequence arrangement module, a condition verification module, a simulation evaluation module, and a state interaction module. The system simulates the production state of new energy power plants to determine a simulation sample set. Based on the simulation density of the simulation sample set, an operating state vector is established, and the operating state vector is arranged in a time series to generate a power generation state sequence. Based on the time interval of the maximum output scenario under the power generation state sequence, simulation conditions for multiple time intervals are obtained, and the performance indicators of each simulation condition are determined. The influence degree of each simulation condition is verified, the goodness of fit of the performance indicators under each simulation condition is analyzed, and the dominant factors of each simulation condition are obtained. Based on the dominant factors of each simulation condition, a simulation state guidance diagram is constructed to determine the simulation control commands. This achieves both accuracy and efficiency in the simulation of new energy power generation.

Description

Technical Field

[0001] This invention relates to the field of power grid technology, specifically a simulation system for new energy power generation. Background Technology

[0002] The large-scale integration of renewable energy generation poses challenges to grid operation. Testing and evaluating the grid connection performance of renewable energy power plants before grid connection is crucial for ensuring the safe and stable operation of the grid. Due to capacity and technological limitations, current testing and field trials can only be conducted on inverters or generation units. The grid connection performance of renewable energy power plants needs to be addressed through simulation. However, existing simulation technologies generally employ single-factor simulations, neglecting factors such as equipment and grid constraints, resulting in reduced accuracy in matching the sample data with real-world scenarios.

[0003] For example, Chinese Patent Publication No. CN112508338A discloses a comprehensive evaluation system for the grid connection performance of new energy power plants, belonging to the field of power grid technology. The system includes a power grid operation data interface for transmitting power grid operation data to a complex power grid operating condition simulation module; a new energy power plant model parameter identification and verification module for model verification using test data; a power grid complex operating condition simulation module for conducting complex power grid simulation; a power grid regional operation risk judgment module for determining operation risks; a new energy power plant grouping module for dividing new energy power plant groups; a support vector machine module for dynamically modeling the grid connection performance indicators of new energy power plants; an indicator correction module for correcting the grid connection performance evaluation indicators of new energy power plants; a comprehensive evaluation module for evaluating grid-connected new energy power plants; and a new energy power plant evaluation result display and output module for displaying and outputting the evaluation results.

[0004] For example, Chinese Patent Publication No. CN113408924A discloses a planning method for a park integrated energy system based on statistical machine learning. The planning method includes: acquiring historical meteorological data, performing power flow calculations on the power generation of distributed photovoltaic power sources, the power generation of the thermal system, and the gas supply power of the natural gas system to obtain the operating state variables of the park integrated energy system, establishing an objective function based on these operating state variables, setting constraints, iteratively solving the planning problem composed of the objective function and constraints, and obtaining a distributed photovoltaic power source site selection and capacity planning scheme.

[0005] Existing technologies focus on ranking and evaluating various performance parameters under grid connection and combining outdoor temperature to predict changes in the energy system temperature. However, these conditions are biased towards human experience settings, which may overlook low-probability, high-risk scenarios. Furthermore, existing technologies prioritize the timeliness of indicators rather than addressing state transitions caused by multiple indicators. This makes it easy to ignore the correlation between multiple factors in complex grid environments, increasing the susceptibility of grid prediction simulations to single-factor influences and ultimately reducing simulation accuracy. This makes it unsuitable for predicting grid changes under the influence of multiple factors. Summary of the Invention

[0006] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is: a new energy power generation production simulation system, comprising: a state simulation module, a sequence arrangement module, a condition verification module, a simulation evaluation module, and a state interaction module; wherein, the output end of the state simulation module is connected to the sequence arrangement module, the output end of the sequence arrangement module is connected to the condition verification module, the output end of the condition verification module is connected to the simulation evaluation module, and the output end of the simulation evaluation module is connected to the state interaction module.

[0007] The state simulation module is used to simulate the production state of each new energy power station using the input data and the simulation density of energy output, and to determine the simulation sample set.

[0008] The sequence arrangement module is used to establish the operating state vector of each simulation operating state according to the simulation density of the simulation sample set, and to arrange the time series in the output form of the operating state vector to generate the power generation state sequence.

[0009] The condition verification module is used to obtain simulation conditions in multiple time intervals based on the time interval of the maximum output scenario under the power generation state sequence; and to determine the performance indicators of each simulation condition based on the energy output density and power generation state of the power generation state sequence.

[0010] The simulation evaluation module is used to verify the degree of influence of each simulation condition on power generation, analyze the goodness of fit of performance indicators under each simulation condition, and obtain the dominant factors of each simulation condition.

[0011] The beneficial effects of this invention are as follows: First, this invention generates a simulation sample set by combining multiple input sources such as meteorological conditions, equipment status, and historical data. The simulation sample set is calculated using a power output probability sequence, and after determining the probability, it is correlated with simulation density to couple the collected data, thereby improving the realism of the generated simulation samples. Then, for the formed operating state vector, constraints are imposed on the power generation state and other aspects based on its state index, power truncation constraints, and rate of change constraints, reducing the data corresponding to power abrupt changes in the simulation sample set, thus ensuring that the equipment of each new energy power station can maintain a safe operating state as the time series progresses.

[0012] Second, this invention sets the parameters that generate the maximum output power by taking the time interval of the maximum output scenario under the power generation state sequence, and processes the simulation conditions by combining its performance in different time intervals. The data of the power generation state sequence is described by indicators such as the average output power and the curtailment rate, so as to generate performance indicators that meet the simulation conditions and adapt to the parameters under different process combinations.

[0013] Third, this invention clusters simulation conditions into single and multiple combinations to determine the degree of influence between simulation conditions under various clusters. It uses transfer entropy to represent different fitting indices in single-class and multi-class clusters to illustrate the data performance of each cluster after clustering. Then, by identifying relatively overlapping parts of the clustered data to obtain the main factors causing the simulation conditions, it uses each factor as a node-guided cyclical combination to explain the control commands required for the current simulation conditions in single-factor to multi-factor scenarios. This enables a closed loop from state analysis to real-time decision-making, assisting maintenance personnel in timely maintenance of the data simulation and improving the environmental adaptability of the simulation. Attached Figure Description

[0014] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0015] Figure 1 This is a schematic diagram of a new energy power generation simulation system.

[0016] Figure 2 This is a flowchart illustrating the state simulation module of a new energy power generation production simulation system.

[0017] Figure 3 This is a flowchart illustrating the sequence layout module of a new energy power generation simulation system.

[0018] Figure 4 This is a flowchart illustrating the condition verification module of a new energy power generation simulation system.

[0019] Figure 5 This is a flowchart illustrating the simulation evaluation module of a new energy power generation simulation system.

[0020] Figure 6 This is a flowchart illustrating the state interaction module of a new energy power generation simulation system. Detailed Implementation

[0021] The embodiments of the present invention are described in detail below. The embodiments described below are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention. Where specific techniques or conditions are not specified in the embodiments, they shall be performed in accordance with the techniques or conditions described in the literature in the art or in accordance with the product manual.

[0022] See Figure 1 A new energy power generation production simulation system includes: a state simulation module, a sequence arrangement module, a condition verification module, a simulation evaluation module, and a state interaction module; wherein, the output of the state simulation module is connected to the sequence arrangement module, the output of the sequence arrangement module is connected to the condition verification module, the output of the condition verification module is connected to the simulation evaluation module, and the output of the simulation evaluation module is connected to the state interaction module.

[0023] The state simulation module is used to simulate the production state of each new energy power station using the input data and the simulation density of energy output, and to determine the simulation sample set.

[0024] The sequence arrangement module is used to establish the operating state vector of each simulation operating state according to the simulation density of the simulation sample set, and to arrange the time series in the output form of the operating state vector to generate the power generation state sequence.

[0025] The condition verification module is used to obtain simulation conditions in multiple time intervals based on the time interval of the maximum output scenario under the power generation state sequence; and to determine the performance indicators of each simulation condition based on the energy output density and power generation state of the power generation state sequence.

[0026] The simulation evaluation module is used to verify the degree of influence of each simulation condition on power generation, analyze the goodness of fit of performance indicators under each simulation condition, and obtain the dominant factors of each simulation condition.

[0027] The state interaction module is used to construct a simulation state guidance diagram based on the dominant factors of each simulation condition, and determine the simulation control commands according to the guidance routes of each simulation condition in the simulation state guidance diagram.

[0028] Preferably, the input data for each new energy power station includes, but is not limited to, the current power generation energy of the new energy power station, historical meteorological data, historical output data, equipment status, and the theoretical output power of the new energy power station. The meteorological data is used to describe the environment in which the current new energy power station generates new energy, and the equipment status is used to describe whether the equipment in the new energy power station can work normally. At the same time, the equipment status will describe the equipment parameters of the new energy power station that generate electricity, so as to determine its theoretical output power, and the relative distribution of historical data, to identify whether the current new energy power station can work normally.

[0029] Preferably, the simulated energy output density uses historical power output data of new energy power plants, and generates its output probability distribution using kernel density estimation. Then, the output probability distribution is used as the simulated energy output density at this time after index extraction.

[0030] Preferred, such as Figure 2 As shown, the implementation of the state simulation module includes: combining the input data of each new energy power station with meteorological conditions, equipment status, and grid conditions to generate simulation samples with different combination structures, and recording the data proportion under each combination structure; the combination structure represents the combination of historical data of the new energy power station with meteorological conditions, equipment status, and grid conditions to illustrate the output power under different conditions, so as to determine the current situation of new energy power generation.

[0031] The probability distribution of the input data of each new energy power station is calculated to generate a power output probability distribution sequence; the simulation density under energy output is extracted according to the power output probability distribution sequence.

[0032] A mapping relationship is constructed between simulation density and simulation samples with different combinations of structures, as well as a mapping relationship between output probability distribution sequence and simulation samples. Based on the two-layer mapping relationship, a simulation sample set is generated.

[0033] For example, taking a scenario with both wind power and photovoltaic power generation as an example, the combination of meteorological conditions will be represented as a combination of three dimensions: wind speed stratification, light intensity stratification, and temperature stratification. The output power of new energy power plants under these data combinations will be classified to find out under what conditions the power generation efficiency and output power are the highest.

[0034] The wind speed stratification can be divided into three levels according to IEC standards: Cut-in wind speed (3-4 m / s), rated wind speed (12-15 m / s), and Cut-out wind speed (25 m / s). This will distinguish and identify the wind speed conditions in wind power generation scenarios and then extract the corresponding output power.

[0035] The light intensity stratification is divided into three levels according to the characteristics of photovoltaic power generation: weak light, medium light, and strong light. The values ​​of 500W / m² and 800W / m² are used as the data to divide the three levels to identify the power generation situation. When only one of photovoltaic power generation and wind power generation exists in the new energy power station, the classification is based on the existence of only one.

[0036] Temperature stratification can be based on ambient temperature or the operating temperature of the equipment, using its average value as the stratification method. Multiple ranges can be set according to the percentage of the operating temperature deviating from the average value in historical data, such as greater than 10%, greater than 30%, and greater than 50% as multiple numerical ranges.

[0037] The equipment status combination links the actual operating status of the equipment with the output current and power in the form of normal operation, derating operation, and fault operation, and is divided into multiple combination structures.

[0038] The grid conditions are combined according to voltage fluctuation, frequency deviation and fault type. Voltage fluctuation is divided into ±5% of rated voltage and ±10% of rated voltage. Frequency deviation refers to the situation at 49.5Hz, 50Hz and 50.5Hz. Fault type can describe three-phase short circuit, single-phase grounding, voltage drop to 0.2pu, etc., to describe the power generation mode of new energy power plants.

[0039] Preferably, the probability distribution calculation of the input data of each new energy power station can be performed using a Gaussian distribution. After calculating the probability value of the output power of the new energy power station, the input data corresponding to each probability value is grouped into an output probability distribution sequence according to the probability value. Then, the index values ​​of the output probability distribution sequence are extracted as the simulation density of energy output, such as the peak value, variance, average value and minimum output power of the output probability distribution sequence, which are used as the simulation density of its output.

[0040] Preferably, the two-layer mapping relationship maps the simulation density according to the values ​​of simulation samples in different combined structures, and according to the probability values ​​of the output probability distribution sequence and the probability of the corresponding simulation sample existing in the input time. This illustrates the distribution mapping of the simulation density as different combined structures are combined, and associates the simulation density with the proportion and combination method of the combined structure. Then, it maps the probability value corresponding to the simulation density with the time associated with the input data. This achieves a mapping method that decouples structure and density and couples probability and behavior to illustrate the complete simulation link from static configuration to dynamic response. The first mapping ensures the combined coverage of the samples, and the second mapping injects temporal probability characteristics, ultimately generating a simulation dataset that combines structural diversity and output realism.

[0041] In one embodiment of the present invention, the simulated operating state refers to the comprehensive operational performance of a new energy power station, such as a wind farm or a photovoltaic power station, at a specific point in time or under specific conditions during the simulation process. It reflects the actual output characteristics of the system under given input conditions, such as wind speed, light intensity, and temperature, and operational constraints, such as equipment capacity and grid dispatch requirements. In wind power simulation, one simulated operating state might have a wind speed of 8 m / s, a wind turbine speed of 12 rpm, an output power of 1.5 MW, and a grid frequency of 50 Hz, while another simulated operating state might have a wind speed of 15 m / s and an output power of 2.0 MW.

[0042] The operating state vector combines multiple sets of power generation data corresponding to the simulation density. For example, the average value and standard deviation corresponding to the simulation density are used as labels, and then the active power, reactive power, wind speed, solar radiation, grid frequency, timestamp and equipment status of each data point are described. This shows the output power and corresponding situation at the simulation density at a specific time point, as shown in Table 1.

[0043] Table 1. Schematic diagram of running state vector

[0044]

[0045] Table 1 illustrates the dimensions included in the operating state vector. These dimensions represent the main parameters involved in power generation at renewable energy plants, and these parameters can be used to assess whether the current renewable energy plant can generate electricity normally. This data is obtained from the input data and arranged in a time series format to obtain a power generation state sequence, simulating continuous changes in actual operation.

[0046] like Figure 3 As shown, the implementation of the sequence arrangement module also includes: establishing a state index between each new energy power station based on the index value corresponding to the simulation density. The state index represents the active power, reactive power, equipment operation status, and environmental factors such as wind speed and sunlight that are directly used for new energy power generation in each energy power station. An index is established for the data of these new energy power stations to facilitate subsequent simulation and processing of the operating status.

[0047] When a status index query is triggered, the operating status of the target device corresponding to each status index is indicated based on the amount of data and the operating status vector. The target device represents the equipment that generates electricity in the new energy power station.

[0048] Based on the operating status of the target equipment, the output format of the operating status vector is determined. The output format is generally divided into numerical, Boolean, and enumerated types. The numerical type directly represents the numerical value of the operating status vector, such as [wind speed = 8.2 m / s, fan speed = 15 rpm, gearbox temperature = 65℃, power generation = 2.5 MW]. This format specifies the exact value of the operating status vector, which is then arranged into a power generation status sequence in time series form. The Boolean type indicates whether the target equipment is functioning normally, using "yes / no" or "1 / 0" to indicate whether the equipment is in a normal state, used for subsequent alarms or logical judgments regarding simulation conditions. The enumerated type describes the mode or category of the equipment's operating status through labels, such as [operating mode = grid-connected power generation, cooling status = automatic, wind direction deviation = left deviation 5°]. These can describe the power generation situation of the equipment in the new energy power station, and the changes in these labels during continuous time sampling are used to identify whether the state of each piece of equipment has changed during new energy power generation.

[0049] Preferably, when the operating state vector is established, the data in its simulation sample set will be standardized, and the data will be normalized to form a dimensionless power generation state sequence for subsequent processing.

[0050] Preferably, when arranging time series data in the form of operating state vector outputs, it is also necessary to process and limit the fluctuations in output power, and limit the difference between predicted and actual power, in order to solve the problems of output fluctuation suppression and curtailment rate optimization in power generation scenarios. At this point, the power generation state sequence of each renewable energy station is adjusted according to its physical constraints, ensuring that the output power value is within the range of minimum fluctuations. This is achieved by processing the operating state vectors of renewable energy stations based on historical data, resulting in a power generation state sequence composed of a time interval, an operating state vector, and an output confidence interval. The output confidence interval refers to the confidence interval corresponding to the output power.

[0051] Preferably, the implementation method of arranging the time series in the form of the operating state vector output includes: setting power truncation constraints and rate of change constraints for the output power of adjacent time points in the operating state vector; the power truncation constraint means that the output power must not exceed the maximum output of the equipment, such as if the rated power of the wind turbine is 2.5MW, then the output power must not exceed this value in the simulation sample set; the rate of change constraint means that the power change at adjacent time points is constrained by its maximum ramp rate, that is, in, This represents the change in power between adjacent time points. This indicates the maximum ramp rate, which is the maximum rate of change in the output power that new energy power generation equipment, such as wind turbines and photovoltaic power stations, can safely adjust per unit time. It represents the time difference between adjacent moments.

[0052] Based on the state index, the operating state vector is introduced into a dual-loop queue containing power truncation constraints and rate of change constraints. When the dual-loop queue points to any state index, the data corresponding to that state index is combined into a power generation state sequence in time series form. This pointing indicates that the operating state vector satisfies both the power truncation and rate of change constraints, and the generated simulation data is more closely related to the real environment. This dynamic adjustment of predicted values ​​ensures that the output power change rate of the new energy power generation equipment remains within a safe range, thereby guaranteeing the stable operation of the power grid and the safe operation of the equipment.

[0053] Preferably, the dual-circular queue represents a queue that retrieves the current operating state vector using power truncation constraints and rate of change constraints. Once all data in the queue satisfies the constraints, the corresponding operating state vectors are arranged into an output power generation state sequence.

[0054] In one embodiment of the present invention, the maximum output scenario refers to the scenario in which the output power of new energy power generation can be maximized under specific simulation conditions. The simulation conditions are a set of parameter combinations designed around the maximum output form to cover the full operating range of the system. The maximum output scenario represents the data corresponding to the maximum output power obtained within a time interval of the power generation state sequence. The parameters of these data are extracted into simulation conditions. Simultaneously, the power generation state sequence contains multiple time intervals to identify the relevant states of the simulation conditions under which different combinations of simulation conditions are applied, thereby selecting the performance indicators for subsequent processing of the simulation conditions.

[0055] like Figure 4 As shown, the implementation of the condition verification module also includes: predicting multiple potential simulation conditions in the current time interval based on the simulation conditions of the previous time interval.

[0056] The potential simulation conditions and simulation conditions of the current time interval are merged to obtain the target simulation conditions of the current time interval.

[0057] The probability of occurrence of the target simulation condition is statistically analyzed. If the probability of occurrence of the target simulation condition is greater than the probability threshold, the corresponding target simulation condition is used as the output simulation condition.

[0058] Preferably, when predicting potential simulation conditions, prediction is based on an LSTM time-series prediction model. The LSTM time-series prediction model uses simulation conditions from the previous time period as input to determine which information to input into the current time interval. For example, it uses a fixed-length window, such as a window length of 6 time steps, corresponding to a 1-hour data segment of historical data, as model input. Each window represents a training sample, labeled with the simulation conditions of the next time step after the end of the window. When the model starts, the hidden states and cell states are initialized, typically set to zero vectors to represent no prior knowledge.

[0059] Then, processing proceeds step by step. The forget gate determines which historical information to discard based on the current input, such as the wind speed in the previous time step and the previous hidden state. For example, if the current wind speed suddenly increases, the forget gate may discard old memories related to the low wind speed.

[0060] The input gate filters the data, determining which new information, such as current wind speed and temperature changes, should be stored in the cell state. For example, the input gate might focus on recording sudden wind speed events while downplaying data with stable changes.

[0061] The cell state integrates forgotten old information and newly added information to generate the current cell state, which is stored as long-term memory. The output gate generates the current hidden state based on the cell state and the current input, which serves as the input for the next time step.

[0062] Finally, the LSTM iterates through the entire historical sequence, progressively updating the hidden states and cell states to generate a context vector representing the comprehensive historical pattern. The cell states of the LSTM are then input into a fully connected layer, mapping them to the potential simulation conditions for the current time interval. The fully connected layer may output three wind speed states with probabilities of 0.6, 0.3, and 0.1, respectively. A Softmax function is then used to ensure that the output probabilities are normalized, reflecting the relative likelihood of each potential condition.

[0063] Because the current simulation conditions represent data obtained from simulating all aspects of the operating state vector, the LSTM will output data including output power and various scenarios. Then, a probability value is assigned to the simulation conditions, calculated using a Gaussian distribution, such as outputting simulation conditions combining "high wind speed (probability 0.6), moderate temperature (probability 0.7), and equipment health (probability 0.85)". As LSTM is a concept already documented in existing technology, it will not be explained further here.

[0064] Preferably, when merging the potential simulation conditions and the existing simulation conditions in the current time interval, the predicted potential simulation conditions are actually Cartesian products of the existing simulation conditions to generate a data set containing all possible combinations; each combination represents a simulation scenario, covering information such as the device environment, and this data set is regarded as the target simulation condition.

[0065] Next, the probability of occurrence of the target simulation condition is calculated. This probability is obtained by solving the conditional probabilities of multiple simulation conditions within the combination corresponding to the target simulation condition. Then, the probability threshold is selected based on the simulation conditions in the combination of the target simulation conditions. For example, the average probability of occurrence of the corresponding simulation condition under the maximum output scenario can be used to set the probability threshold. Alternatively, a threshold that covers historical scenarios as much as possible can be selected. For example, the occurrence probabilities of historical simulation conditions can be sorted, and the minimum occurrence probability corresponding to the cumulative probability reaching 90% can be used as the probability of the current target simulation condition to filter simulation conditions that are as close as possible to historical scenarios.

[0066] Preferably, taking wind power and photovoltaic power generation as examples, energy output density can represent the power generation per unit area of ​​photovoltaic panels within a certain period of time, or it can represent the actual power generation per unit area of ​​wind turbine swept area within a certain period of time. This certain period of time is in the form of days or years, to illustrate the power generation of new energy power generation under the metering method per unit area under the simulation conditions corresponding to the power generation state sequence.

[0067] The generation state represents the standard deviation of the output power of the generation state sequence and the curtailment rate. The curtailment rate indicates the proportion of unused renewable energy due to conservative forecasting or equipment limitations; this part represents the ratio of unused available power to output power under the current simulation conditions, illustrating the underutilization of the generation state sequence. The curtailment rate is calculated by subtracting the output power in the current sequence from the theoretical power, and then dividing the difference by the theoretical power. The energy output density and the corresponding generation state values ​​are then used as performance indicators for the simulation conditions.

[0068] In one embodiment of the present invention, when verifying the impact of each simulation condition on power generation, the simulation conditions are first divided into simulation conditions under multiple similar scenarios. Then, the mapping relationship between each simulation condition and the performance index after fitting is statistically analyzed to find out how the simulation conditions affect the output power of the power generation. Finally, the degree of influence and main factors between the simulation conditions are output to describe the impact of the simulation index on the power generation environment. If the simulation conditions are used as input conditions, cluster correction is performed to evaluate the fitting form of the performance index of each simulation condition under the simulation and the concurrent form of its performance index, so as to determine the impact of each simulation condition on the simulation process.

[0069] like Figure 5As shown, the implementation of the simulation evaluation module also includes: taking each simulation condition as input, determining a single-class cluster for a single simulation condition and a multi-class cluster for combinations of multiple simulation conditions; when setting the classification clusters, using DBSCAN to cluster each simulation condition, classifying the content of a single simulation condition and combinations of multiple simulation conditions to obtain a set of multiple similar scenarios, such as classifying actual power generation scenarios such as "high wind speed and low irradiance". A single simulation condition represents the set of parameters that reach maximum output power within a time period, illustrating which similar scenarios exist for the scenario corresponding to this parameter set, facilitating subsequent analysis of the conditions that lead to the maximum output power in that scenario; the set of combinations of multiple simulation conditions is more inclined to illustrate how the conditions for reaching maximum output power in the next time interval are transformed when the maximum output power is reached in one time interval, used to explain whether there are corresponding causal relationships under multiple scenarios, leading to state transitions and other content related to the existence of multiple simulation conditions. In essence, a single-class cluster is a set obtained by clustering the parameter set corresponding to a single simulation condition to identify simulation conditions similar to that single simulation condition. A multi-class cluster, on the other hand, combines multiple simulation conditions and identifies a set of scenarios similar to these combined simulation conditions. For example, in some cases, a single-class cluster could be a wind speed parameter cluster, while a multi-class cluster could be a wind speed-light intensity combined cluster, thus illustrating the impact of the acquired simulation conditions on the output power.

[0070] Based on the obtained multi-class and single-class clusters, the influence degree of each simulation condition is set according to the output power values ​​of each simulation condition within the multi-class and single-class clusters.

[0071] Preferably, the degree of influence of each simulation condition can be represented by the ratio of the mean output power in a single-class cluster to the standard deviation in the corresponding multi-class cluster. A higher ratio indicates that the output power is relatively more concentrated or deviates less from the overall standard deviation under that condition, suggesting better stability or less impact from the condition. Conversely, a lower ratio indicates a greater impact on output power, leading to increased data dispersion. By comparing the degree of influence corresponding to each simulation condition, it is possible to identify which parameter set has a more significant impact and pinpoint the low-ratio components, which may represent key sensitive points in system performance, thus indicating the stability of the current equipment's power generation.

[0072] As shown in Table 2, the content corresponding to the current single-class cluster is compared with the corresponding data of the multi-class cluster to determine the relationship between the data corresponding to each parameter under different combinations of simulation conditions in different scenarios, and to obtain the goodness of fit corresponding to the simulation conditions.

[0073] Table 2. Classification Cluster Diagram

[0074]

[0075] If sample i1 exists in the single-class cluster under the current simulation conditions, then the silhouette coefficient within the single-class cluster can be expressed as: : ;in, This represents the average distance from sample i1 to other samples in the current single-class cluster. This distance is calculated based on the performance indicators under simulation conditions. When calculating these performance indicators, since they are expressed as the standard deviation of power generation, curtailment rate, and output power, the samples representing the parameter set under simulation conditions are normalized with other samples in the current single-class cluster in vector form. The distance value is then obtained using Euclidean distance calculation. ;in, This represents the number of samples in the current single-class cluster, with i1 and j1 ranging from 1 to n1. and These represent the normalized values ​​of the energy output density for samples i1 and j1, respectively. and These represent the normalized standard deviations of the output power of samples i1 and j1, respectively. and These represent the normalized values ​​of the curtailment rate for samples i1 and j1, respectively. This represents the average distance from sample i1 to the nearest sample in the single-class cluster. This distance value is calculated in the same way as the distance above, so it will not be described again here. The larger the value, the higher the density and the better the separation of the single-class clusters combined by a single simulation condition. This is to illustrate the relationship between the output power represented by these performance indicators after using simulation condition clustering.

[0076] The conditional probability of its multi-class cluster can be calculated based on the formula for calculating conditional probability, using the currently sampled simulation data as a basis.

[0077] The transfer entropy represents the transition from a parameter corresponding to one simulation condition to a parameter corresponding to another simulation condition, which is the state based on a combination of multiple simulation conditions. The state transitions to another set of multiple simulation condition combinations. The transfer entropy That is, the transfer entropy at this time. ;in, Representing state and state Conditional probability, Representing state The probability of a state transition can be calculated based on a Gaussian distribution. It represents the relationship between two multi-class clusters by expressing the conditional probability of a state transitioning from one value to another, using the performance index of the parameter set after multiple simulation conditions. The ranges of i2 and j2 represent the number of multi-class clusters where state transitions occur.

[0078] After completing the goodness-of-fit processing of its performance indicators, the dominant factors of its simulation conditions are identified based on the goodness-of-fit. That is, the implementation of the simulation evaluation module also includes: judging the goodness-of-fit of the performance indicators corresponding to each simulation condition under single-class cluster and multi-class cluster respectively, sorting the influence degree and goodness-of-fit of each simulation condition by feature, and taking the sorted output as the dominant factor of each simulation condition.

[0079] Preferably, when sorting features, for a single-class cluster, if the influence and goodness of fit of a certain simulation condition are the highest, then this highest value can be taken as the dominant factor of that simulation condition. For example, if the profile coefficient of wind speed is the highest, it is a single dominant factor affecting the output power of the current power generation. If the transfer entropy of wind speed plus grid frequency is the highest in a multi-class cluster, then the combination of multiple simulation conditions is taken as its dominant factor. Therefore, it is necessary to perform multi-objective processing on the output content of single-class and multi-class clusters to obtain the most relevant set of dominant factors under the current simulation condition, so as to guide the subsequent description of the power system state transition, thereby analyzing the final form of change trend of the output power of the current new energy power station under the current data setting.

[0080] The implementation methods for ranking the influence and goodness of fit of each simulation condition include: for the influence of simulation conditions in single-class and multi-class clusters, obtaining the duration, output power and parameter range of each simulation condition in single-class and multi-class clusters, and determining the key features of the current simulation condition and related simulation conditions.

[0081] For each key feature, identify overlapping data within each time interval to form an overlapping dataset.

[0082] The overlapping datasets are compared with the associated simulation conditions, and the data parameter with the largest goodness of fit after comparison is selected as the dominant factor in the output.

[0083] Preferably, the key features of the simulation conditions are extracted based on the values ​​of duration, output power, and parameter range within single-classification clusters and multi-classification clusters. For example, the key features are selected from the data corresponding to the simulation conditions with the highest average output power or the longest duration in single-classification clusters; the associated simulation conditions are selected from the data of multiple simulation condition combinations corresponding to the highest transition probability or the highest output power in multi-classification clusters.

[0084] Next, the key features are extracted from the data points that are repeated in each time region. These data represent the general situation of power generation by the equipment of the new energy power station. Then, the overlapping datasets are compared with the associated simulation conditions to capture the synergistic effect of single and multiple factors. Then, the parts that are the same when comparing the key features with the associated simulation conditions are processed, and the goodness of fit of the simulation conditions corresponding to the same parts is selected. Since the goodness of fit of the simulation conditions is divided into multiple simulation condition combinations and single simulation conditions, the goodness of fit identified at this time will select the part corresponding to the maximum value of the two to output, so as to explain the main data that is mainly affected at this time. For example, in one case, the goodness of fit calculated for the wind speed range of the single-class cluster is 0.85, and the goodness of fit calculated for the wind speed range plus the light range of the multi-class cluster is 0.92. In this case, the range of the corresponding parameters in the multi-class cluster will be selected as the dominant factor for subsequent simulation to explain the main factors affecting the new energy power generation under the current situation.

[0085] In one embodiment of the present invention, such as Figure 6 As shown, the implementation of the state interaction module also includes: using each dominant factor as a node, and using the degree of influence of the simulation conditions corresponding to the dominant factor as the weight of the edges between the nodes to form a simulation state guidance graph.

[0086] Based on the goodness of fit of each node in the simulation state guidance diagram when connecting, the guidance of the simulation state guidance diagram is set, and the shortest connection path in each guidance is used as the guidance route for each simulation condition.

[0087] Based on the output power on the guide path under each simulation condition, a threshold triggering strategy is set, and simulation control commands are selected according to the threshold triggering strategy.

[0088] Preferably, the threshold triggering strategy is set by the average value of the data corresponding to each simulation condition in historical data, or by the average value of the time interval closest to the simulation condition being processed; for example, when the output power of the corresponding node on the conductor route is greater than or less than the threshold, different simulation control commands are selected.

[0089] Preferably, the direction of the simulation state guidance diagram is determined based on the transfer entropy calculated by the simulation conditions in the multi-class cluster. If the transfer entropy from the simulation conditions corresponding to the current dominant factor to the simulation conditions of another dominant factor is greater than zero, then the direction is considered to be from the current dominant factor to another dominant factor; otherwise, the direction is the opposite.

[0090] Preferably, the shortest connection path in each guide is determined by the shortest path algorithm, which uses the values ​​of each dominant factor plus the weight of each node to determine the shortest path after connecting each node with the guide.

[0091] Preferably, the implementation method of selecting simulation control commands according to the threshold triggering strategy further includes: for each simulation condition, extracting the sub-paths of each guidance route, and superimposing the output power of each sub-path with the output power of the corresponding guidance route to obtain the path power of each guidance route.

[0092] Threshold retrieval is performed based on the path power of each guide route to identify the threshold triggering strategy for the current guide route; when the path power of the guide route is greater than the threshold of the threshold triggering strategy, the simulation control command is selected based on the minimum difference in the output power of the guide routes within adjacent time intervals.

[0093] When the path power of the guide route is less than the threshold of the threshold triggering strategy, the simulation control command is selected according to the simulation conditions corresponding to the maximum output power in the current guide route.

[0094] The sub-paths of each of the aforementioned guiding routes are obtained by dividing the guiding route under each simulation condition into multiple indivisible sub-paths, summing the output power along each sub-path, and then searching the database for simulation control commands based on the path power value. When the power exceeds a threshold, the approach of minimizing power fluctuations is selected, i.e., the control command corresponding to the smallest difference in the database is chosen to avoid equipment overload or grid instability caused by sudden power changes. If the power falls below the threshold triggering strategy, the path power needs to be quickly increased, for example, by prioritizing the processing of the node corresponding to the highest output power to quickly restore its power generation efficiency.

[0095] For example, the simulation control instruction for minimum difference selection gradually reduces the power of the current path to prevent overloading of equipment such as wind turbines. The simulation control instruction corresponding to the maximum output power adjusts the equipment's operating parameters to increase the overall output power, facilitating subsequent intervention by maintenance personnel. Furthermore, after outputting the simulation control instruction, the data regulated by the instruction is used again in the state simulation module to achieve a logical control method of simulation → verification → simulation. This allows for multiple simulation adjustments during production simulation, improving the accuracy of the simulation output data.

[0096] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention, which are still covered within the protection scope of the present invention.

Claims

1. A simulation system for new energy power generation, characterized in that, include: The state simulation module is used to simulate the production state of each new energy power station based on the input data and the simulation density of energy output, and to determine the simulation sample set. The sequence arrangement module is used to establish the operating state vector of each simulation operating state according to the simulation density of the simulation sample set, and to arrange the time series in the output form of the operating state vector to generate the power generation state sequence. The condition verification module is used to obtain simulation conditions in multiple time intervals based on the time interval of the maximum output scenario under the power generation state sequence; and to determine the performance indicators of each simulation condition based on the energy output density and power generation state of the power generation state sequence. The simulation evaluation module is used to verify the degree of influence of each simulation condition on power generation, analyze the goodness of fit of performance indicators under each simulation condition, and obtain the dominant factors of each simulation condition. The state interaction module is used to construct a simulation state guidance diagram based on the dominant factors of each simulation condition, and determine the simulation control commands according to the guidance path of each simulation condition in the simulation state guidance diagram. The implementation methods for the state interaction module also include: Using each dominant factor as a node, and the degree of influence of the dominant factor on the simulation conditions as the weight of the edges between nodes, a simulation state guidance graph is formed. Based on the goodness of fit of each node in the simulation state guidance diagram when connecting, the guidance of the simulation state guidance diagram is set, and the shortest connection path in each guidance is used as the guidance route for each simulation condition. Based on the output power on the guide path under each simulation condition, a threshold triggering strategy is set, and simulation control commands are selected according to the threshold triggering strategy.

2. The new energy power generation simulation system according to claim 1, characterized in that, The input data for new energy power stations include, but are not limited to, the current power generation energy of the new energy power station, historical meteorological data, historical output data, equipment status, and the theoretical output power of the new energy power station.

3. The new energy power generation simulation system according to claim 1, characterized in that, The state simulation module can be implemented in the following ways: The input data of each new energy power station are combined using meteorological conditions, equipment status, and power grid conditions to generate simulation samples with different combination structures, and the data proportion under each combination structure is recorded. The probability distribution of the input data of each new energy power station is calculated to generate a power output probability distribution sequence; the simulation density under energy output is extracted according to the power output probability distribution sequence. A mapping relationship is constructed between simulation density and simulation samples with different combinations of structures, as well as a mapping relationship between output probability distribution sequence and simulation samples. Based on the two-layer mapping relationship, a simulation sample set is generated.

4. The new energy power generation simulation system according to claim 1, characterized in that, The implementation methods of the sequence arrangement module also include: Based on the index values ​​corresponding to the simulation density, a state index is established between each new energy power station. When a status index query is triggered, the running status of the target device corresponding to each status index is indicated based on the amount of data corresponding to each status index and the running status vector. The output form of the operating state vector is determined based on the operating state of the target device.

5. The new energy power generation simulation system according to claim 4, characterized in that, The implementation methods for arranging time series data in the form of running state vector outputs include: For the output power at adjacent moments in the operating state vector, power truncation constraints and rate of change constraints are set sequentially; Based on the state index, the operating state vector is introduced into a double-circular queue of power cutoff constraints and rate of change constraints. When the double-circular queue points to any state index, the data corresponding to the state index is combined into a power generation state sequence in the form of a time series.

6. The new energy power generation simulation system according to claim 1, characterized in that, The implementation methods of the condition validation module also include: Based on the simulation conditions of the previous time interval, predict multiple potential simulation conditions for the current time interval. The potential simulation conditions and simulation conditions of the current time interval are merged to obtain the target simulation conditions of the current time interval; The probability of occurrence of the target simulation condition is statistically analyzed. If the probability of occurrence of the target simulation condition is greater than the probability threshold, the corresponding target simulation condition is used as the output simulation condition.

7. The new energy power generation simulation system according to claim 1, characterized in that, The simulation evaluation module can also be implemented in the following ways: Using each simulation condition as input, determine the single-class cluster under a single simulation condition and the multi-class cluster under a combination of multiple simulation conditions; Based on the obtained multi-class clusters and single-class clusters, the influence degree of each simulation condition is set according to the output power values ​​of each simulation condition within the multi-class clusters and single-class clusters. The goodness of fit of the performance indexes corresponding to each simulation condition under single-class cluster and multi-class cluster are determined respectively. The influence degree and goodness of fit of each simulation condition are ranked by feature, and the ranked output results are used as the dominant factors of each simulation condition.

8. The new energy power generation simulation system according to claim 7, characterized in that, The methods for ranking the influence and goodness of fit of various simulation conditions include: To assess the impact of simulation conditions within single-class and multi-class clusters, the duration, output power, and parameter range of each simulation condition within single-class and multi-class clusters are obtained, and the key characteristics of the current simulation condition and related simulation conditions are determined. For each key feature, identify overlapping data within each time interval to form an overlapping dataset; The overlapping datasets are compared with the associated simulation conditions, and the data parameter with the largest goodness of fit after comparison is selected as the dominant factor in the output.

9. The new energy power generation simulation system according to claim 1, characterized in that, The implementation methods for selecting simulation control instructions based on the threshold triggering strategy also include: For each simulation condition, the sub-paths of each guide route are extracted, and the output power of each sub-path is superimposed with the output power of the corresponding guide route to obtain the path power of each guide route. Threshold retrieval is performed based on the path power of each guide route to identify the threshold triggering strategy for the current guide route; when the path power of the guide route is greater than the threshold of the threshold triggering strategy, the simulation control command is selected based on the minimum difference in the output power of the guide routes within adjacent time intervals. When the path power of the guide route is less than the threshold of the threshold triggering strategy, the simulation control command is selected according to the simulation conditions corresponding to the maximum output power in the current guide route.

Images