Energy storage power station potential safety hazard analysis method and system based on digital twinning

Abstract

The invention relates to the technical field of energy storage power station management, and discloses an energy storage power station potential safety hazard analysis method and system based on digital twinning. And a three-dimensional geometric model of the energy storage power station is constructed by using a digital twin model construction module, so that visual mapping of data is realized. The data calculation module calculates the spatial gradient change trend of each index, and the potential safety hazard prediction module predicts the potential safety hazard level in combination with a deep learning algorithm. The early warning module automatically triggers and sends early warning information according to a prediction result, and the user interaction module provides a visual display interface. The system can monitor the operation state of the energy storage power station in real time, accurately predict potential safety hazards, timely give out early warning, effectively improve the safety and stability of the energy storage power station, reduce the risk of safety accidents, and provide powerful guarantee for stable operation of a power system.

Description

Technical Field

[0001] The present invention relates to the technical field of energy storage power station management, and specifically provides a method and system for analyzing potential safety hazards of an energy storage power station based on digital twin. Background Technique

[0002] With the transformation of the energy structure and the rapid development of renewable energy, energy storage power stations, as a key link in energy storage and conversion, are crucial for the stable operation of the power system. However, during operation, energy storage power stations face various potential safety hazards, such as excessive temperature, abnormal current and voltage, and excessive concentration of harmful gases. If these problems cannot be detected and addressed in a timely manner, they are likely to trigger safety accidents and cause serious damage to personnel and equipment. With the rapid development of technologies such as the Internet of Things, big data, and artificial intelligence, digital twin technology, as an emerging digital means, provides a new solution for analyzing potential safety hazards of energy storage power stations.

[0003] Digital twin technology realizes deep interaction and integration between the physical world and the digital world by constructing a virtual model that corresponds one-to-one with the physical energy storage power station. However, there are still many deficiencies in the application of existing digital twin technology in analyzing potential safety hazards of energy storage power stations. On the one hand, the means of data collection and processing are not intelligent enough, making it difficult to accurately monitor and effectively manage key parameters such as temperature, current and voltage, and concentration of harmful gases. On the other hand, there is a lack of efficient data mapping and visualization means, making it difficult for operation and maintenance personnel to intuitively understand the operating status and distribution of potential safety hazards of the energy storage power station.

[0004] Most existing potential safety hazard prediction models are based on simple threshold judgment or statistical methods, making it difficult to accurately predict the potential safety hazard level and potential risks of energy storage power stations. Therefore, there is an urgent need for an intelligent analysis system that can monitor the operating status of energy storage power stations in real time, accurately predict the potential safety hazard level, and issue early warnings in a timely manner, so as to improve the safety and stability of energy storage power stations and reduce the probability of safety accidents. Summary of the Invention

[0005] The purpose of the present invention is to provide a method and system for analyzing potential safety hazards of an energy storage power station based on digital twin to solve the problems raised in the above background technique.

[0006] To achieve the above purpose, the present invention provides the following technical solution: An energy storage power station safety hazard analysis system based on digital twin, the system includes;

[0007] A data collection module, including a temperature sensing unit, a current and voltage monitoring unit, and a gas concentration detection unit, for regularly collecting temperature data, current and voltage data, and harmful gas concentration data in the energy storage power station;

[0008] A digital twin model construction module for constructing a digital twin model of an energy storage power station based on the collected data. The model includes a visual representation of the distribution of temperature data, current and voltage data, and harmful gas concentration in the energy storage power station; the digital twin model construction module further includes:

[0009] A data receiving unit for receiving the temperature data, current and voltage data, and harmful gas concentration data in the energy storage power station transmitted by the data acquisition module, and these data are represented in the form of spatial distribution;

[0010] A data preprocessing unit for cleaning, standardizing, and formatting the received data;

[0011] A 3D modeling unit that uses a 3D modeling algorithm based on geometric modeling and physical modeling to construct a 3D geometric model of the energy storage power station according to the actual layout, equipment dimensions, and parameter information of the energy storage power station;

[0012] A data mapping unit. For temperature data, it uses a linear interpolation algorithm to map temperature values to the color space, and represents the temperature level through the change of color hue; for current and voltage data, it uses a threshold segmentation algorithm to map different ranges of current and voltage values to different textures; for harmful gas concentration data, it uses a grading method to map it to a point cloud map for representation, and selects different color modes to represent the concentration distribution; using the 3D rendering engine OpenGL, it renders the mapped 3D geometric model and the data on it to generate a digital twin model of the energy storage power station;

[0013] A data calculation module for calculating the gradient change trend of temperature, current and voltage, and harmful gas concentration in the spatial coordinates;

[0014] A safety hazard prediction module for constructing a safety hazard prediction model. The safety hazard prediction model uses the gradient change trend of each index output by the data calculation module and combines a deep learning algorithm to predict the safety hazard level of the energy storage power station;

[0015] An early warning module that automatically triggers and sends corresponding early warning information according to the safety hazard level obtained by the safety hazard prediction module;

[0016] A user interaction module for displaying the digital twin model, safety hazard prediction results, and early warning information, and receiving the operation instructions of the user.

[0017] Preferably, the calculation method of the data calculation module is:

[0018] For any point (x, y, z) in the three-dimensional space, its index gradient is calculated by the following formula:

[0019]

[0020] Among them, are the partial derivatives of the index L in the x, y, and z directions, respectively.

[0021] Preferably, the data mapping unit:

[0022] For temperature data, a linear interpolation algorithm is used to map the temperature value to the color space. The specific calculation formula is:

[0023]

[0024] Among them, C(T) is the color value corresponding to the temperature T, C min and C max are the minimum and maximum values of the predefined color range respectively, T min and T max are the minimum and maximum values of the temperature data respectively;

[0025] For current and voltage data, a threshold segmentation algorithm is used to map different ranges of current and voltage values to different textures. Specifically, a series of thresholds V 1 , V 2 ,..., V n are set. According to the interval to which the current and voltage value V belongs, the corresponding texture is selected for mapping;

[0026] For harmful gas concentration data, a grading method is used to map it to a point cloud map for representation. Specifically, a series of concentration levels L 1 , L 2 ,..., L m are set. Each level corresponds to a color pattern. According to the level to which the harmful gas concentration value belongs, the corresponding color pattern is selected for representation.

[0027] Preferably, the safety hazard prediction model is constructed using the convolutional neural network CNN algorithm.

[0028] Preferably, the structure of the safety hazard prediction model includes:

[0029] An input layer, which is used to receive the gradient change trend data of each index output by the data calculation module, including temperature gradient, current and voltage gradient, and harmful gas concentration gradient. The data is input in the form of a three-dimensional matrix;

[0030] Multiple convolutional layers, each convolutional layer contains multiple convolutional kernels, which are used to extract the local features of the input data. The convolution operation is carried out through the following formula:

[0031]

[0032] Among them, is the output of the j-th convolutional kernel in the l-th layer, is the output of the (l - 1)-th layer, is the weight of the j-th convolutional kernel in the l-th layer corresponding to the i-th input feature, is the bias term, f() is the activation function, M j represents the set of input features connected to the j-th convolutional kernel;

[0033] The pooling layer is used to reduce the feature dimension of the convolutional layer output, reduce the computational amount, and retain important features at the same time;

[0034] The fully connected layer is used to map the features output by the pooling layer to the safety hazard level space, and calculate the prediction probabilities of each safety hazard level through the softmax function.

[0035] Preferably, the training steps of the safety hazard prediction model include:

[0036] Step A: Data preparation, dividing the gradient change trend data of each index output by the data calculation module into a training set and a test set;

[0037] Step B: Initialize the model parameters, including the weights and bias terms of the convolutional kernels;

[0038] Step C: Forward propagation, input the training data into the model, and calculate the prediction results of the output layer;

[0039] Step D: Calculate the loss, and use the cross-entropy loss function to calculate the error between the prediction result and the true safety hazard level; the calculation formula of the cross-entropy loss function is:

[0040]

[0041] where N is the number of samples, y i is the true label of the i-th sample, p i is the prediction probability of the i-th sample;

[0042] Step E: Backward propagation, calculate the gradient according to the loss function, and update the model parameters by the gradient descent method;

[0043] Step F: Repeat steps C to E until the loss function converges or reaches the preset number of training epochs.

[0044] Preferably, the implementation method of the warning module includes:

[0045] Set warning thresholds, and preset warning thresholds corresponding to different safety hazard levels according to the safety standards and operation requirements of the energy storage power station, including low-risk, medium-risk, high-risk, and extremely high-risk levels;

[0046] Receive the safety hazard level and obtain real-time safety hazard level information from the safety hazard prediction module;

[0047] Early warning judgment: Compare the obtained safety hazard level with the preset early warning threshold to determine whether an early warning needs to be triggered;

[0048] Early warning information generation: Generate corresponding early warning information according to the judgment result. The early warning information includes the safety hazard level, early warning time, early warning reason, and recommended measures to be taken.

[0049] Preferably, the data preprocessing unit cleans the received data in ways including missing value processing and outlier processing:

[0050] Methods for processing missing values include:

[0051] For numerical features, fill in the missing values with the median of the non-missing values of the feature;

[0052] For categorical features, fill in the missing values with the mode of the non-missing values of the feature;

[0053] For missing values in time series data, fill them with the average of the adjacent values before and after;

[0054] For features with obvious business meanings and few missing values, fill them manually according to business logic;

[0055] Methods for processing outliers include:

[0056] For numerical features, use the 3σ principle to identify and process outliers; specifically: calculate the mean and standard deviation of the feature, and regard the values exceeding the mean ± 3 times the standard deviation as outliers.

[0057] Preferably, the method for the data preprocessing unit to standardize the data is:

[0058] Use the Z-score method to subtract each numerical data from its mean and divide by its standard deviation, so that the processed data conforms to the standard normal distribution. The formula for standardization is: Z = (X - μ) / σ, where X is the original data, μ is the mean of the original data, σ is the standard deviation of the original data, and Z is the standardized data.

[0059] Preferably, a method for analyzing safety hazards of an energy storage power station based on digital twin, the method includes the following steps:

[0060] S1. Data collection step: Regularly collect temperature data, current and voltage data, and harmful gas concentration data in the energy storage power station through a temperature sensing unit, a current and voltage monitoring unit, and a gas concentration detection unit;

[0061] S2. Steps for constructing the digital twin model: S201. Receive and preprocess the collected data; S202. Based on the 3D modeling algorithm of geometric modeling and physical modeling, combined with the actual layout, equipment dimensions and parameter information of the energy storage power station, construct a 3D geometric model of the energy storage power station; S203. Map the temperature data to the color space through the linear interpolation algorithm, map the current and voltage data to different textures through the threshold segmentation algorithm, map the harmful gas concentration data to the point cloud map through the grading method, and use the OpenGL rendering engine to generate the digital twin model of the energy storage power station;

[0062] S3. Data calculation steps: Calculate the gradient change trend of temperature, current and voltage, and harmful gas concentration in the spatial coordinates;

[0063] S4. Safety hazard prediction steps: Construct a safety hazard prediction model, and use the calculated gradient change trend of each index, combined with the deep learning algorithm, to predict the safety hazard level of the energy storage power station;

[0064] S5. Warning steps: According to the safety hazard level obtained from the safety hazard prediction steps, automatically trigger the warning module to generate and send corresponding warning information;

[0065] S6. User interaction steps: Display the digital twin model, safety hazard prediction results and warning information through the user interaction module, and receive and process the user's operation instructions.

[0066] Compared with the prior art, the beneficial effects of the present invention are:

[0067] By integrating the temperature sensing unit, current and voltage monitoring unit and gas concentration detection unit, the present invention realizes the comprehensive, real-time and high-precision collection of key parameters in the energy storage power station. The data preprocessing unit cleans, standardizes and formats the collected data, effectively eliminates abnormal data, improves the accuracy and reliability of the data, and provides a solid foundation for the subsequent construction of the digital twin model and safety hazard prediction.

[0068] The present invention adopts data mapping technology to map key parameters such as temperature, current and voltage, and harmful gas concentration into the digital twin model of the energy storage power station in the form of intuitive colors, textures and point cloud maps, realizing the visual display of data. Through the rendering of the 3D rendering engine OpenGL, the operation and maintenance personnel can intuitively observe the operation status and safety hazard distribution of the energy storage power station, greatly improving the monitoring efficiency and accuracy.

[0069] The present invention constructs a safety hazard prediction model based on deep learning algorithms. This model can accurately predict the safety hazard levels of energy storage power stations by using the trend of gradient changes of various indicators output by the data calculation module. The warning module automatically triggers and sends corresponding warning information according to the safety hazard prediction results, timely reminding the operation and maintenance personnel to take measures, and effectively reducing the probability of safety accidents. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] Figure 1 is the working principle diagram of a safety hazard analysis system for energy storage power stations based on digital twin according to the present invention;

[0071] Figure 2 is the working principle diagram of the data mapping unit;

[0072] Figure 3 is the training flow chart of the safety hazard prediction model;

[0073] Figure 4 is the processing flow chart of the warning module. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0074] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0075] Please refer to Figures 1-4 , the present invention provides a technical solution: a safety hazard analysis system for energy storage power stations based on digital twin, the system includes:

[0076] A data acquisition module, including a temperature sensing unit, a current and voltage monitoring unit, and a gas concentration detection unit, is used to regularly collect temperature data, current and voltage data, and harmful gas concentration data in the energy storage power station; the collected data is transmitted to the data receiving unit through a communication interface.

[0077] A digital twin model construction module is used to construct a digital twin model of the energy storage power station according to the collected data. This model includes a visual representation of the distribution of temperature data, current and voltage data, and harmful gas concentration in the energy storage power station; the digital twin model construction module further includes:

[0078] A data receiving unit, which receives various data in the energy storage power station transmitted by the data acquisition module. These data are represented in the form of spatial distribution, including the specific values of temperature, current and voltage, and harmful gas concentration and their corresponding position information.

[0079] The data preprocessing unit cleans the received data; then performs standardization processing to ensure the comparability of data from different sources and with different dimensions; finally, performs formatting processing to convert the data into a format suitable for subsequent processing. The methods for cleaning the received data include missing value processing and outlier processing:

[0080] The methods for processing missing values include:

[0081] For numerical features, fill the missing values with the median of the non-missing values of the feature;

[0082] For categorical features, fill the missing values with the mode of the non-missing values of the feature;

[0083] For the missing values in time series data, fill them with the average of the adjacent values before and after;

[0084] For features with obvious business meanings and few missing values, fill them manually according to business logic;

[0085] The methods for processing outliers include:

[0086] For numerical features, use the 3σ principle to identify and process outliers; specifically: calculate the mean and standard deviation of the feature, and regard the values exceeding the mean ± 3 times the standard deviation as outliers.

[0087] The method for standardizing the data is:

[0088] Use the Z-score method to subtract the mean of each numerical data from it and divide by its standard deviation, so that the processed data conforms to the standard normal distribution. The formula for standardization processing is: Z = (X - μ) / σ, where X is the original data, μ is the mean of the original data, σ is the standard deviation of the original data, and Z is the standardized data.

[0089] The 3D modeling unit uses a 3D modeling algorithm based on geometric modeling and physical modeling to construct an accurate 3D geometric model according to the actual layout, equipment dimensions and parameter information of the energy storage power station. The model includes detailed information such as the building structure, equipment layout, and pipeline routing of the energy storage power station.

[0090] The data mapping unit, for temperature data, uses a linear interpolation algorithm to map the temperature values to the color space, and represents the temperature level through the change of color hue; for current and voltage data, uses a threshold segmentation algorithm to map different ranges of current and voltage values to different textures; for harmful gas concentration data, uses a grading method to map it to a point cloud map for representation, and selects different color modes to represent the concentration distribution; uses the 3D rendering engine OpenGL to render the mapped 3D geometric model and the data on it to generate a digital twin model of the energy storage power station.

[0091] The data calculation module is responsible for calculating the gradient change trends of temperature, current voltage, and harmful gas concentration in the spatial coordinates. By performing spatial interpolation and gradient calculation on the collected data, the change trends of each index in space are obtained, providing basic data for safety hazard prediction.

[0092] The safety hazard prediction module constructs a safety hazard prediction model. This model uses the gradient change trends of each index output by the data calculation module and combines deep learning algorithms to predict the safety hazards of the energy storage power station. By training the model, it can identify the patterns and trends that may lead to safety hazards and predict the safety hazard levels of the energy storage power station.

[0093] The warning module automatically triggers and sends corresponding warning messages according to the safety hazard levels obtained by the safety hazard prediction module. The warning messages can remind users of the safety status of the energy storage power station through various means such as sound, light, and electricity, and prompt them to take corresponding measures for intervention.

[0094] The user interaction module provides a friendly user interface for displaying the digital twin model, safety hazard prediction results, and warning messages. Users can view the real-time status of the energy storage power station through this module, understand the prediction and warning of safety hazards, and issue operation instructions as needed. At the same time, this module also supports functions such as data query and report generation for users to facilitate decision-making and analysis of the safety management of the energy storage power station.

[0095] The present invention will be further described below in conjunction with Embodiments 1 to 3:

[0096] Embodiment 1:

[0097] The mapping of temperature data uses a linear interpolation algorithm to convert the temperature values into color representations in the color space for intuitive display of the temperature distribution in the energy storage power station. The specific steps are as follows:

[0098] Set the color range: Select a color gradient scheme, such as from blue (representing low temperature) to red (representing high temperature), and determine the minimum value C min and the maximum value C max .

[0099] Obtain the temperature data range: Obtain the minimum value T min and the maximum value T max of the temperature data in the energy storage power station from the data acquisition module.

[0100] Apply the linear interpolation algorithm: For each temperature value T, calculate its corresponding color value C(T) using the following formula:

[0101]

[0102] For example, if T min = 20 °C, T max = 50 °C, C min is blue, and C max is red, then when the temperature T = 35 °C, the corresponding color can be obtained by calculation as blue-violet, indicating a medium temperature.

[0103] The mapping of current-voltage data uses a threshold segmentation algorithm to divide the current-voltage values into different intervals, and each interval corresponds to a texture to distinguish different states of the current-voltage. The specific steps are as follows:

[0104] Set thresholds: According to the safe operation standards of the energy storage power station, a series of current-voltage thresholds V 1 , V 2 ,..., V n are set to divide the current-voltage values into n + 1 intervals.

[0105] Texture selection: Specify a texture for each interval. For example, use a green texture for the low-voltage interval, a yellow texture for the normal-voltage interval, a red texture for the high-voltage interval, etc.

[0106] Map textures: According to the interval to which the current-voltage value V belongs, select the corresponding texture for mapping. For example, if V is between V 2 and V 3 , then it is mapped to a yellow texture.

[0107] The mapping of harmful gas concentration data uses a grading method to divide the concentration values into different grades, and each grade corresponds to a color pattern to visually display the distribution of harmful gases. The specific steps are as follows:

[0108] Set concentration grades: According to the safe concentration standards of harmful gases, a series of concentration grades L 1 , L 2 ,..., L m are set, and each grade corresponds to a color pattern. For example, use a green point cloud for low concentration, a yellow point cloud for medium concentration, and a red point cloud for high concentration.

[0109] Map color patterns: According to the grade to which the harmful gas concentration value belongs, select the corresponding color pattern for representation. For example, if the concentration value is between L 2 and L 3 , then it is mapped to a yellow point cloud.

[0110] Using the 3D rendering engine OpenGL, render the mapped 3D geometric model and the data thereon to generate a digital twin model of the energy storage power station. During the rendering process, temperature data is represented on the model surface through color hue changes, current and voltage data is represented on key parts of the model through different textures, and harmful gas concentration data is represented in the space around the model through different color modes of the point cloud map. In this way, users can intuitively understand the distribution of temperature, current, voltage, and harmful gas concentration inside the energy storage power station through the digital twin model, providing strong support for the prediction and early warning of potential safety hazards.

[0111] The calculation method of the said data calculation module is as follows: For any point (x, y, z) in the 3D space, its index gradient is calculated through the following formula:

[0112]

[0113] where, are the partial derivatives of the index L in the x, y, and z directions respectively, The indices represented by represent the change gradients of temperature, current, voltage, and harmful gas concentration respectively.

[0114] Embodiment 2:

[0115] The potential safety hazard prediction module predicts the potential safety hazard level of the energy storage power station by constructing and training a Convolutional Neural Network (CNN) model. The specific structure and training steps of the potential safety hazard prediction model will be elaborated in detail below.

[0116] The structure design of the potential safety hazard prediction model is as follows:

[0117] Input layer: Receive the safety index gradient data in the form of a 3D matrix output by the data calculation module, including temperature gradient, current and voltage gradient, and harmful gas concentration gradient. Assuming that the data dimension of each index is 64×64×1, the data dimension of the input layer is 64×64×3, corresponding to the gradient maps of the three safety indices respectively.

[0118] Convolutional layer: The model contains two convolutional layers. Each convolutional layer uses 32 convolutional kernels of size 3×3, with a stride of 1 and a padding method of'same' to keep the feature map size unchanged. After the convolutional operation, introduce non-linearity through the ReLU activation function. The formula is as follows:

[0119]

[0120] where, is the output of the jth convolutional kernel in the lth layer, is the output of the (l - 1)th layer, is the weight corresponding to the i-th input feature for the j-th convolutional kernel in the l-th layer, is the bias term, f() is the activation function, M j represents the set of input features connected to the j-th convolutional kernel.

[0121] Pooling layer: After each convolutional layer, a max-pooling layer is connected. The pooling window size is 2×2 and the stride is 2 to reduce the dimension of the feature map and retain important features.

[0122] Fully connected layer: After the pooling layer, the feature map is flattened into a one-dimensional vector and input into two fully connected layers. The first fully connected layer has 128 neurons, and the second fully connected layer has the same number of neurons as the number of safety hazard levels (assumed to be 5 levels). The softmax function is used to output the predicted probability for each level.

[0123] Output layer: The output layer directly gives the predicted result of the safety hazard level, that is, the level with the highest probability.

[0124] The training steps of the safety hazard prediction model include:

[0125] Data preparation: Obtain historical data from the data calculation module, including a three-dimensional matrix of temperature gradient, current-voltage gradient, and concentration gradient of harmful gases, as well as the corresponding safety hazard level labels. Randomly divide the data into a training set (80%) and a test set (20%).

[0126] Model initialization: Use a random initialization method (such as He initialization) to assign values to the weights of the convolutional kernel and the fully connected layer, and initialize the bias term to 0.

[0127] Forward propagation: Input the training data into the model, and calculate the predicted probability distribution of the safety hazard level through the convolutional layer, pooling layer, fully connected layer, and softmax layer.

[0128] Loss calculation: Use the cross-entropy loss function to calculate the error between the predicted result and the true label. The formula is as follows:

[0129]

[0130] where N is the number of samples, y i is the true label of the i-th sample, and p i is the predicted probability of the i-th sample.

[0131] Backward propagation and parameter update: Use the Adam optimizer to update the model parameters (weights and biases) through the backward propagation algorithm according to the gradient of the loss function. The Adam optimizer combines the advantages of the momentum method and the RMSprop algorithm, can adaptively adjust the learning rate, and accelerate the convergence speed.

[0132] Iterative training: Repeatedly execute the forward propagation, loss calculation, and backpropagation steps until the loss function converges (i.e., the loss value no longer decreases significantly) or reaches a preset number of training epochs (e.g., 100 epochs).

[0133] Evaluate the performance of the model on the test set, and calculate metrics such as accuracy, precision, recall, and F1-score to verify the generalization ability and prediction accuracy of the model.

[0134] Example 3:

[0135] The task of the warning module is to automatically determine whether a warning needs to be triggered based on the safety hazard level output by the safety hazard prediction module and send the corresponding warning information. The following are the specific implementation steps of the warning module:

[0136] According to the safety management specifications and actual operation requirements of the energy storage power station, set the warning thresholds corresponding to different safety hazard levels. For example, the safety hazard levels can be divided into four levels: low risk, medium risk, high risk, and extremely high risk, and the warning thresholds are set correspondingly. The low-risk level may indicate a relatively small safety hazard and only requires attention; the medium-risk level may indicate a relatively large safety hazard and measures need to be taken for monitoring; the high-risk level may indicate a serious safety hazard and immediate measures need to be taken for handling; the extremely high-risk level indicates an extremely serious safety hazard that may endanger the safe operation of the energy storage power station and requires an emergency response.

[0137] The warning module receives the safety hazard level information from the safety hazard prediction module in real time. The safety hazard prediction module analyzes the trend of gradient changes of various indicators of the energy storage power station based on a convolutional neural network (CNN), predicts the current safety hazard level, and transmits this level information to the warning module.

[0138] The warning module compares the received safety hazard level with the preset warning threshold. For example, if the level output by the safety hazard prediction module is high risk, and the warning threshold corresponding to the high-risk level has been set to require triggering a warning, the warning module determines that a warning needs to be issued.

[0139] When the warning is judged to need to be triggered, the warning module generates the corresponding warning information. The warning information includes the safety hazard level, warning time, warning reason, and recommended measures, etc. For example, the warning information may contain content such as "Safety hazard level: high risk; Warning time: XX:XX, XX / XX / XXXX; Warning reason: The temperature gradient has increased abnormally, which may cause equipment overheating; Recommended measure: Immediately check and reduce the equipment temperature".

[0140] The warning module sends warning messages to relevant personnel or systems through preset communication methods. The communication methods can include text messages, emails, system message pushes, etc. For example, the warning module can send warning messages to the operation and maintenance personnel of the energy storage power station in the form of text messages, send emails to the power station management, and notify relevant monitoring systems through system message pushes.

[0141] After each warning is triggered, the warning module records the warning message and its handling situation in the warning log. The warning log includes fields such as warning time, safety hazard level, warning message content, receiving personnel or systems, handling situation, etc. Through the warning log, historical warning information can be conveniently queried and analyzed, providing data support for the safety management of the energy storage power station.

[0142] The present invention also includes a method for analyzing safety hazards of an energy storage power station based on digital twin, and the method includes the following steps:

[0143] S1. Data collection step: Regularly collect temperature data, current and voltage data, and harmful gas concentration data in the energy storage power station through a temperature sensing unit, a current and voltage monitoring unit, and a gas concentration detection unit;

[0144] S2. Digital twin model construction step: S201. Receive and preprocess the collected data; S202. Based on a three-dimensional modeling algorithm of geometric modeling and physical modeling, combined with the actual layout, equipment dimensions, and parameter information of the energy storage power station, construct a three-dimensional geometric model of the energy storage power station; S203. Map the temperature data to the color space through a linear interpolation algorithm, map the current and voltage data to different textures through a threshold segmentation algorithm, map the harmful gas concentration data to a point cloud map through a grading method, and use the OpenGL rendering engine to generate a digital twin model of the energy storage power station;

[0145] S3. Data calculation step: Calculate the gradient change trends of temperature, current and voltage, and harmful gas concentration in the spatial coordinates;

[0146] S4. Safety hazard prediction step: Construct a safety hazard prediction model, and use the calculated gradient change trends of each index, combined with a deep learning algorithm, to predict the safety hazard level of the energy storage power station;

[0147] S5. Warning step: According to the safety hazard level obtained in the safety hazard prediction step, automatically trigger the warning module to generate and send corresponding warning messages;

[0148] S6. User interaction step: Display the digital twin model, safety hazard prediction results, and warning messages through the user interaction module, and receive and process the operation instructions of the user.

[0149] The implementation manner of this method refers to the above-mentioned embodiments and will not be elaborated in the specification.

[0150] It should be noted that, in this document, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprising", "including" or any other variant thereof are intended to cover non-exclusive inclusion, such that a process, method, article or device comprising a series of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article or device.

[0151] Although the embodiments of the present invention have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention, and the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A safety hazard analysis system for energy storage power stations based on digital twins, characterized in that: The system comprises: The data acquisition module includes a temperature sensing unit, a current and voltage monitoring unit, and a gas concentration detection unit, which are used to regularly collect temperature data, current and voltage data, and harmful gas concentration data in the energy storage power station; A digital twin model construction module is used to construct a digital twin model of the energy storage power station based on the collected data, and the model includes a visual representation of temperature data, current and voltage data, and the distribution of harmful gas concentrations in the energy storage power station; the digital twin model construction module further includes: A data receiving unit is used to receive temperature data, current and voltage data, and harmful gas concentration data in the energy storage power station transmitted by the data acquisition module, and these data are represented in the form of spatial distribution; A data pre-processing unit is used to clean, standardize and format the received data; The three-dimensional modeling unit uses a three-dimensional modeling algorithm based on geometric modeling and physical modeling to construct a three-dimensional geometric model of the energy storage power station according to the actual layout, equipment size and parameter information of the energy storage power station; The data mapping unit uses a linear interpolation algorithm to map temperature data to the color space, and uses color hue changes to indicate the temperature. For current and voltage data, a threshold segmentation algorithm is used to map current and voltage values ​​in different ranges to different textures. For harmful gas concentration data, a hierarchical classification method is used to map it to a point cloud map, and different color modes are selected to indicate the distribution of concentration. The three-dimensional rendering engine OpenGL is used to render the mapped three-dimensional geometric model and the data on it to generate a digital twin model of the energy storage power station. Data calculation module, used to calculate the gradient change trend of temperature, current voltage and harmful gas concentration in spatial coordinates; A safety hazard prediction module is used to construct a safety hazard prediction model. The safety hazard prediction model uses the gradient change trend of each indicator output by the data calculation module and combines it with a deep learning algorithm to predict the safety hazard level of the energy storage power station; The early warning module automatically triggers and sends corresponding early warning information according to the safety hazard level obtained by the safety hazard prediction module; The user interaction module is used to display the digital twin model, safety hazard prediction results and warning information, and receive user operation instructions.

2. According to the digital twin-based energy storage power station safety hazard analysis system of claim 1, it is characterized in that: The calculation method of the data calculation module is: For any point (x, y, z) in three-dimensional space, its index gradient Calculated by the following formula: in, and are the partial derivatives of the index L in the x, y and z directions respectively.

3. According to the digital twin-based energy storage power station safety hazard analysis system of claim 1, it is characterized in that: The data mapping unit: For temperature data, a linear interpolation algorithm is used to map the temperature value to the color space. The specific calculation formula is: Among them, C(T) is the color value corresponding to temperature T, C min and C max are the minimum and maximum values ​​of the predefined color range, T min and T max are the minimum and maximum values ​​of the temperature data respectively; For current and voltage data, a threshold segmentation algorithm is used to correspond current and voltage values ​​in different ranges to different textures. Specifically, a series of thresholds V1, V2, ..., V n , according to the interval to which the current and voltage value V belongs, the corresponding texture is selected for mapping; For the concentration data of harmful gases, the level division method is adopted to map it to the point cloud representation, specifically setting a series of concentration levels L1, L2, ..., L m Each level corresponds to a color mode. According to the level to which the harmful gas concentration value belongs, the corresponding color mode is selected for representation.

4. According to the digital twin-based energy storage power station safety hazard analysis system of claim 1, it is characterized in that: The safety hazard prediction model is constructed using a convolutional neural network (CNN) algorithm.

5. According to claim 4, a digital twin-based energy storage power station safety hazard analysis system is characterized in that: The structure of the potential safety hazard prediction model includes: The input layer is used to receive the gradient change trend data of each indicator output by the data calculation module, including temperature gradient, current voltage gradient and harmful gas concentration gradient. The data is input in the form of a three-dimensional matrix; Multiple convolutional layers, each of which contains multiple convolution kernels, are used to extract local features of the input data. The convolution operation is performed using the following formula: in, is the output of the jth convolution kernel in the lth layer, is the output of the l-1th layer, is the weight of the j-th convolution kernel in the l-th layer corresponding to the i-th input feature, is the bias term, f() is the activation function, M j Represents the input feature set connected to the jth convolution kernel; The pooling layer is used to reduce the feature dimension of the convolutional layer output, reduce the amount of calculation, and retain important features; The fully connected layer is used to map the features output by the pooling layer to the safety hazard level space and calculate the predicted probability of each safety hazard level through the softmax function.

6. The energy storage power station safety hazard analysis system based on digital twin according to claim 5 is characterized in that: The training steps of the potential safety hazard prediction model include: Step A: Data preparation, dividing the gradient change trend data of each indicator output by the data calculation module into a training set and a test set; Step B: Initialize the model parameters, including the weights and bias terms of the convolution kernel; Step C: Forward propagation, input the training data into the model and calculate the prediction results of the output layer; Step D: Calculate the loss, using the cross entropy loss function to calculate the error between the prediction result and the actual safety hazard level; the calculation formula of the cross entropy loss function is: Where N is the number of samples, y i is the true label of the i-th sample, p i is the predicted probability of the i-th sample; Step E: Back propagation, calculate the gradient according to the loss function, and update the model parameters by gradient descent method; Step F: Repeat steps C to E until the loss function converges or the preset number of training rounds is reached.

7. The energy storage power station safety hazard analysis system based on digital twin according to claim 1 is characterized in that: The implementation of the early warning module includes: Set warning thresholds. According to the safety standards and operation requirements of the energy storage power station, pre-set warning thresholds corresponding to different safety hazard levels, including low risk, medium risk, high risk and extreme risk levels; Receive the safety hazard level and obtain real-time safety hazard level information from the safety hazard prediction module; Early warning judgment: compare the acquired safety hazard level with the preset early warning threshold to determine whether an early warning needs to be triggered; Early warning information is generated based on the judgment results. The corresponding early warning information includes the safety hazard level, warning time, warning reason and recommended measures.

8. The energy storage power station safety hazard analysis system based on digital twin according to claim 1 is characterized in that: The data preprocessing unit cleans the received data in the following ways: Methods for handling missing values ​​include: For numerical features, the median of the non-missing values ​​of the feature is used to fill the missing values; For categorical features, the missing values ​​are filled with the mode of the non-missing values ​​of the feature; For missing values ​​in time series data, the average value of the adjacent values ​​is used to fill them; For features with obvious business meaning and few missing values, they are manually filled according to business logic; Methods for dealing with outliers include: For numerical features, the 3σ principle is used to identify and process outliers; specifically, the mean and standard deviation of the feature are calculated, and values ​​exceeding the mean ± 3 times the standard deviation are considered outliers.

9. The energy storage power station safety hazard analysis system based on digital twin according to claim 1 is characterized in that: The method for the data preprocessing unit to perform standardization processing on the data is: The Z-score method is used to subtract the mean of each numerical data and divide it by its standard deviation so that the processed data conforms to the standard normal distribution. The formula for standardization is: Z = (X-μ) / σ, where X is the original data, μ is the mean of the original data, σ is the standard deviation of the original data, and Z is the standardized data.

10. A method for analyzing safety hazards of energy storage power stations based on digital twins, characterized in that: The method comprises the following steps: S1, data collection step: regularly collect temperature data, current and voltage data and harmful gas concentration data in the energy storage power station through the temperature sensing unit, current and voltage monitoring unit and gas concentration detection unit; S2. Digital twin model construction steps: S201. Receive and preprocess the collected data; S202. Based on the three-dimensional modeling algorithm of geometric modeling and physical modeling, the three-dimensional geometric model of the energy storage power station is constructed in combination with the actual layout, equipment size and parameter information of the energy storage power station; S203. The temperature data is mapped to the color space through the linear interpolation algorithm, the current and voltage data are mapped to different textures through the threshold segmentation algorithm, the harmful gas concentration data is mapped to the point cloud map through the level division method, and the digital twin model of the energy storage power station is generated using the OpenGL rendering engine; S3, data calculation step: calculating the gradient change trend of temperature, current voltage and harmful gas concentration in the spatial coordinates; S4, safety hazard prediction step: construct a safety hazard prediction model, use the calculated gradient change trend of each indicator, combined with the deep learning algorithm, to predict the safety hazard level of the energy storage power station; S5, early warning step: according to the safety hazard level obtained in the safety hazard prediction step, the early warning module is automatically triggered to generate and send corresponding early warning information; S6. User interaction step: Display the digital twin model, safety hazard prediction results and warning information through the user interaction module, and receive and process user operation instructions.

Images