Security risk assessment method and system for power generation enterprise

Abstract

The invention discloses a safety risk assessment method and system for a power generation enterprise, and relates to the technical field of enterprise safety assessment, and the method comprises the steps: collecting the equipment data and environment data of the power generation enterprise through a sensor, and carrying out the data preprocessing; performing batch and hierarchical analysis on the preprocessed data, extracting multi-dimensional features, and forming a comprehensive feature set through feature optimization; respectively processing the comprehensive feature set by using a linear model and a nonlinear model, and outputting a prediction result; and carrying out dynamic combination on linear and nonlinear prediction results, and outputting a final risk assessment result. According to the invention, through the multi-level data processing and dynamic combination model, the linear and non-linear relationship in the power generation enterprise equipment and environment data can be comprehensively and accurately captured, so that the accurate assessment of the equipment safety risk is realized. Compared with a traditional method, the accuracy and the real-time performance of risk assessment are greatly improved.

Description

Technical Field

[0001] The present invention relates to the technical field of enterprise safety assessment, and in particular to a safety risk assessment method and system for power generation enterprises. Background Art

[0002] With the expansion of the scale of power generation enterprises and the increase in the complexity of equipment, the safe operation of equipment is crucial to the economic and social benefits of enterprises. Traditional safety risk assessment methods mainly rely on regular inspections and empirical judgments. However, this method is not only highly subjective, but also fails to detect potential risks in a timely manner, which can easily lead to equipment failures or major accidents. Especially in modern power generation enterprises, the equipment operating environment is complex, and various factors (such as temperature, vibration, current, etc.) may affect the operating status of the equipment. How to comprehensively and accurately assess the safety risks of equipment has become an urgent problem that power generation enterprises need to solve.

[0003] Most of the existing security risk assessment systems on the market rely on a single algorithm model and fail to fully utilize the complex relationships in equipment and environmental data. These models can only handle linear relationships or simple nonlinear relationships and have difficulty capturing complex interactions between data. In addition, these systems usually lack the ability to dynamically adjust and cannot adaptively optimize the assessment model according to actual operating conditions, resulting in insufficient accuracy and robustness of the assessment results.

[0004] In order to solve these problems, the present invention proposes a safety risk assessment method for power generation enterprises, which aims to comprehensively capture the linear and nonlinear relationships in equipment and environmental data through multi-level data processing and dynamic combination models, thereby achieving accurate assessment of equipment safety risks. Summary of the invention

[0005] In view of the above-mentioned problems, the present invention is proposed.

[0006] Therefore, the technical problem solved by the present invention is: how to comprehensively utilize linear and nonlinear models based on the complex data of the power generation enterprise's equipment and environment, dynamically adjust and optimize the evaluation model, and evaluate the safety risks of the equipment.

[0007] In order to solve the above technical problems, the present invention provides the following technical solutions: a safety risk assessment method for power generation enterprises, which comprises the following steps:

[0008] Use sensors to collect power generation enterprise equipment data and environmental data and perform data preprocessing;

[0009] The preprocessed data is analyzed in batches and layers to extract multi-dimensional features, and a comprehensive feature set is formed through feature optimization;

[0010] Use linear and nonlinear models to process the comprehensive feature set and output the prediction results;

[0011] Dynamically combine linear and nonlinear prediction results to output the final risk assessment results.

[0012] As a preferred solution of the safety risk assessment method for power generation enterprises described in the present invention, the power generation enterprise equipment data and environmental data include voltage data, current data, power data, speed data, temperature rise data, winding temperature data, load data, cooling water temperature data and cooling water pressure data, external environment temperature data, humidity data, wind speed data and air pressure data;

[0013] The data preprocessing is to detect missing values ​​on the collected raw data, fill in the missing values ​​with interpolation, remove noise in the data with the Kalman filter method, calculate the standard score of the data with the Z-score method, mark the data points exceeding the set threshold as abnormal data, detect and process the abnormal values ​​with the IQR method, and eliminate the abnormal data; normalize the data of different dimensions, convert the data to a unified numerical range, and classify and batch process the normalized data according to the equipment category and time period.

[0014] As a preferred solution of the safety risk assessment method for power generation enterprises described in the present invention, wherein: the comprehensive feature set is batch processing of the pre-processed data;

[0015] Process the equipment data of different categories in batches according to the equipment categories, divide the data according to time periods, extract basic statistical features from the batch-processed data, perform dynamic feature extraction based on the basic statistical features, calculate the change slope, trend direction and fluctuation amplitude of the data, and identify dynamic change features;

[0016] Advanced feature extraction is performed based on the results of dynamic feature extraction, and the periodic features, modal features and nonlinear relationship features of the data are extracted through nonlinear analysis methods;

[0017] Optimize based on the results of advanced feature extraction, remove redundant features through importance analysis and correlation screening, and generate a comprehensive feature set;

[0018] The extracting of basic statistical features includes calculating a weighted mean, a standard deviation, a maximum value, a minimum value and a corrected variance.

[0019] As a preferred solution of the safety risk assessment method for power generation enterprises described in the present invention, the dynamic feature extraction calculates the change slope, trend direction and fluctuation amplitude of the data, identifies the dynamic change characteristics, and the expression is:

[0020]

[0021] Amplitude t =ΔX·(1+skew(X))

[0022]

[0023]

[0024] Among them, β t is the slope of change of trend adjustment, t i is the i-th time point, is the mean value at a time point, is the adjustment coefficient of the standard deviation, Amplitude t is the fluctuation range, ΔX is the extreme value difference, skew(X) is the skewness of the data, AR(X) is the autoregressive model, φ0, φ k is the autoregressive coefficient, the weight in the autoregressive model, X t-k is the data value lagged k periods later, ε t is the noise term, Trend AR It is the direction of autoregressive trend. When φ1>0, the current data is positively correlated with the data at the previous time point. When φ1<0, the current data is negatively correlated with the data at the previous time point. When φ1=0, there is no correlation between the current data and the data at the previous time point.

[0025] The periodic characteristics, modal characteristics and nonlinear relationship characteristics of the data are extracted by nonlinear analysis method, and the expression is:

[0026]

[0027]

[0028]

[0029] Where P(f) is the energy spectral density of the periodic characteristic, λ is the attenuation factor, and f k is the kth frequency component, f k-1 is the k-1th frequency component, β t is the slope of change after trend adjustment, M(t) is the modal characteristic, a n is the amplitude of the nth mode, f n is the frequency of the nth mode, φ n is the phase of the nth mode, g(t) is the attenuation function, Trend AR is the trend direction, K(X,Y) is the dynamic kernel function, which represents the nonlinear relationship between the two data sets, and α i , α jis the characteristic coefficient, ||X i -Y j || 2 is the Euclidean distance, which represents the distance between two data points, σ ​​is the kernel width, which controls the smoothness of the kernel function, and Amplitude t is the fluctuation range;

[0030] The redundant features are removed through importance analysis and correlation screening to generate a comprehensive feature set, which is expressed as:

[0031]

[0032]

[0033]

[0034] FeatureSet opt ={f select ∣Importance(f i )>θ}

[0035] Among them, Importance(f i ) is the contribution of the feature to the result, Var adj (f i ) is the corrected variance of the i-th feature, ∑Var adj (f j ) is the sum of the corrected variances of all features, β t is the slope of change after trend adjustment, The feature f i and f j The correlation coefficient between the two features indicates the linear relationship between the two features. Cov(f i ,f j ) is the feature f i and f j The covariance between The features f i and f j The standard deviation of δ is the correlation threshold, f select FeatureSet is the filtered feature set. opt is a comprehensive feature set, θ is the importance threshold, and important features are screened.

[0036] As a preferred solution of the safety risk assessment method for power generation enterprises described in the present invention, the use of linear and nonlinear models to process the comprehensive feature set respectively includes using the least squares method to find the intercept in the linear regression model, and then processing the comprehensive feature set through the linear regression model to generate a linear prediction output, the expression is:

[0037]

[0038] Among them, β0 is the intercept of the model, β i is the i-th feature f i The regression coefficient, f i is the i-th feature in the comprehensive feature set, γ j is the coefficient of the interaction term between features, indicating that two features f i and f j The impact of the interaction between linear is the error term of the linear model;

[0039] The neural network model is used to process the comprehensive feature set to generate nonlinear prediction output, which is expressed as:

[0040] y nonlinear =NN(F)

[0041] Among them, NN(F) is the neural network model, and F is the overall input of the high-level optimization feature set.

[0042] As a preferred solution of the safety risk assessment method for power generation enterprises described in the present invention, the dynamic combination of linear and nonlinear prediction results is to dynamically combine the linear prediction output and the nonlinear prediction output through a nonlinear integration function, and introduce the interaction terms between the features, and dynamically adjust the weights and interaction terms in the combination function after training, and the expression is:

[0043]

[0044] Among them, y final is the final prediction output, g(·) is the nonlinear integration function, h(y linear ,y nonlinear ) is the combination function, λ ij is the coefficient of the feature interaction term, f i ·f j The feature f i and f j The interaction terms between .

[0045] As a preferred embodiment of the safety risk assessment method for power generation enterprises described in the present invention, the risk assessment result is a comparison between the result of dynamic combination output based on linear prediction output and nonlinear prediction output and a threshold value. If the result of the dynamic combination output is greater than the threshold value, it is determined to be a dangerous state. If the result of the dynamic combination output is less than or equal to the threshold value, it is determined to be a safe state.

[0046] Another object of the present invention is to provide a safety risk assessment system for power generation enterprises, which can capture the linear and nonlinear relationships in equipment and environmental data through multi-level data processing and dynamic combination models, achieve accurate assessment of equipment safety risks, and effectively process complex equipment operation data.

[0047] In order to solve the above technical problems, the present invention provides the following technical solutions: A safety risk assessment system for power generation enterprises, comprising: a data acquisition and preprocessing module, a feature extraction and optimization module, a linear and nonlinear modeling module, a combination and integration module, and a risk assessment and decision support module;

[0048] The data acquisition and preprocessing module collects data from various equipment and environmental sensors of power generation enterprises in real time, and cleans, normalizes and denoises the collected raw data;

[0049] The feature extraction and optimization module extracts important features from the preprocessed data, optimizes the extracted features by using importance analysis and correlation screening, removes redundant features, and generates a comprehensive feature set;

[0050] The linear and nonlinear modeling module uses a linear regression model to model the feature set, captures the linear relationship in the data and the linear interaction between the features, and uses a neural network model and a nonlinear model to process the nonlinear relationship in the data;

[0051] The combination and integration module integrates the output results of the linear and nonlinear models, uses a nonlinear combination function to dynamically adjust the impact of the two models, and automatically adjusts the weights and interaction terms in the combination function through training to optimize the final risk assessment output;

[0052] The risk assessment and decision support module determines the current safety status of the system based on the final output of the combination and integration module, compares the predicted value with the preset threshold, and evaluates whether it is within the safety range.

[0053] A computer device comprises a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the above-mentioned safety risk assessment method for power generation enterprises when executing the computer program.

[0054] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of a safety risk assessment method for a power generation enterprise as described above.

[0055] Beneficial effects of the invention: The invention can comprehensively and accurately capture the linear and nonlinear relationships in the equipment and environmental data of power generation enterprises through multi-level data processing and dynamic combination models, thereby realizing accurate assessment of equipment safety risks. Compared with traditional methods, the invention greatly improves the accuracy and real-time performance of risk assessment. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. Among them:

[0057] Figure 1 An overall flow chart of a safety risk assessment method for power generation enterprises provided by the first embodiment of the present invention;

[0058] Figure 2 A structural diagram of a safety risk assessment system for power generation enterprises provided in accordance with a second embodiment of the present invention. DETAILED DESCRIPTION

[0059] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0060] Example 1, reference Figure 1 As an embodiment of the present invention, a safety risk assessment method for a power generation enterprise is provided, comprising:

[0061] Use sensors to collect power generation enterprise equipment data and environmental data and perform data preprocessing;

[0062] The preprocessed data is analyzed in batches and layers to extract multi-dimensional features, and a comprehensive feature set is formed through feature optimization;

[0063] Use linear and nonlinear models to process the comprehensive feature set and output the prediction results;

[0064] Dynamically combine linear and nonlinear prediction results to output the final risk assessment results.

[0065] The equipment data and environmental data of power generation enterprises include voltage data, current data, power data, speed data, temperature rise data, winding temperature data, load data, cooling water temperature data and cooling water pressure data, external environment temperature data, humidity data, wind speed data and air pressure data;

[0066] Data preprocessing is to detect missing values ​​in the collected raw data, fill in the missing values ​​with interpolation, remove noise in the data with the Kalman filter method, calculate the standard score of the data with the Z-score method, mark the data points exceeding the set threshold as abnormal data, detect and process the abnormal values ​​with the IQR method, and eliminate the abnormal data; normalize the data of different dimensions, convert the data to a unified numerical range, and classify and batch process the normalized data according to the equipment category and time period.

[0067] Process the preprocessed data in batches;

[0068] The equipment data of different categories are processed in batches according to the equipment category, and the data is divided according to the time period. The basic statistical features are extracted from the batch processed data. Based on the basic statistical features, dynamic feature extraction is performed to calculate the change slope, trend direction and fluctuation amplitude of the data, and the dynamic change features are identified. The expression is:

[0069]

[0070] Amplitude t =ΔX·(1+skew(X))

[0071]

[0072]

[0073] Among them, β t is the slope of change of trend adjustment, t i is the i-th time point, is the mean value at a time point, is the adjustment coefficient of the standard deviation, Amplitude t is the fluctuation range, ΔX is the extreme value difference, skew(X) is the skewness of the data, AR(X) is the autoregressive model, φ0, φ k is the autoregressive coefficient, the weight in the autoregressive model, X t-k is the data value lagged k periods later, ε t is the noise term, Trend AR It is the direction of the autoregressive trend. When φ1>0, the current data is positively correlated with the data at the previous time point. When φ1<0, the current data is negatively correlated with the data at the previous time point. When φ1=0, there is no correlation between the current data and the data at the previous time point.

[0074] Advanced feature extraction is performed based on the results of dynamic feature extraction. The periodic features, modal features and nonlinear relationship features of the data are extracted through nonlinear analysis methods. The expression is:

[0075]

[0076]

[0077]

[0078] Where P(f) is the energy spectral density of the periodic characteristic, λ is the attenuation factor, and f k is the kth frequency component, f k-1 is the k-1th frequency component, β t is the slope of change after trend adjustment, M(t) is the modal characteristic, a n is the amplitude of the nth mode, f n is the frequency of the nth mode, φ n is the phase of the nth mode, g(t) is the attenuation function, Trend AR is the trend direction, K(X,Y) is the dynamic kernel function, which represents the nonlinear relationship between the two data sets, and α i , α j is the characteristic coefficient, ||X i -Y j || 2 is the Euclidean distance, which represents the distance between two data points, σ ​​is the kernel width, which controls the smoothness of the kernel function, and Amplitude t The fluctuation range.

[0079] Based on the results of advanced feature extraction, we optimize and remove redundant features through importance analysis and correlation screening to generate a comprehensive feature set. The expression is:

[0080]

[0081]

[0082]

[0083] FeatureSet opt ={f select ∣Importance(f i )>θ}

[0084] Among them, Importance(f i ) is the contribution of the feature to the result, Var adj (f i) is the corrected variance of the i-th feature, ∑Var adj (f j ) is the sum of the corrected variances of all features, β t is the slope of change after trend adjustment, ρ ij The feature f i and f j The correlation coefficient between the two features indicates the linear relationship between the two features. Cov(f i ,f j ) is the feature f i and f j The covariance between The features f i and f j The standard deviation of δ is the correlation threshold, f select FeatureSet is the filtered feature set. opt is a comprehensive feature set, θ is the importance threshold, and important features are screened.

[0085] It is further explained that:

[0086] The entire feature extraction and optimization process mainly captures and processes the nonlinear features in the data, while retaining the analysis of linear relationships. The formula itself does not directly distinguish between linear and nonlinear features, but processes and optimizes these features separately through different analysis methods, and finally generates an advanced optimization feature set.

[0087] Ultimately, these features will be brought into the subsequent security risk assessment model, and the choice of model will depend on whether the overall performance of these features is more linear or nonlinear. This is why in practical applications, it is necessary to judge the linear and nonlinear characteristics of the data and select the appropriate model.

[0088] In the early feature extraction process, data was processed through methods such as dynamic feature extraction and nonlinear analysis. The data relationships captured by these methods may be either linear or nonlinear.

[0089] Therefore, the final generated advanced optimization feature set contains a variety of different types of features, including features obtained through linear analysis methods (such as change slope and fluctuation amplitude) and features obtained through nonlinear analysis methods (such as periodic features and modal features). These features together constitute a complete feature set for subsequent safety risk assessment models.

[0090] The extracting of basic statistical features includes calculating a weighted mean, a standard deviation, a maximum value, a minimum value and a corrected variance.

[0091] The intercept in the linear regression model is obtained using the least squares method, and then the comprehensive feature set is processed through the linear regression model to generate a linear prediction output, which is expressed as:

[0092]

[0093] Among them, β0 is the intercept of the model, β i is the i-th feature f i The regression coefficient, f i is the i-th feature in the comprehensive feature set, γ j is the coefficient of the interaction term between features, indicating that two features f i and f j The impact of the interaction between linear is the error term of the linear model;

[0094] The neural network model is used to process the comprehensive feature set to generate nonlinear prediction output, which is expressed as:

[0095] y nonlinear =NN(F)

[0096] Among them, NN(F) is the neural network model, and F is the overall input of the high-level optimization feature set.

[0097] The linear prediction output and the nonlinear prediction output are dynamically combined through the nonlinear integration function, and the interaction terms between the features are introduced. The weights and interaction terms in the combination function are dynamically adjusted after training. The expression is:

[0098]

[0099] Among them, y final is the final prediction output, g(·) is the nonlinear integration function, h(y linear ,y nonlinear ) is the combination function, λ ij is the coefficient of the feature interaction term, f i ·f j The feature f i and f j The interaction terms between .

[0100] It is further explained that

[0101] h(y linear ,y nonlinear )=σ(α1·y linear +α2·y nonlinear +α3·y linear ·y nonlinear )

[0102] Among them, σ(·) is the Sigmoid activation function used to introduce nonlinear transformation to ensure that the output is between 0 and 1, which is suitable for risk scoring scenarios. α1 and α2 are weights automatically optimized through the model training process, which control the contribution of linear and nonlinear models respectively. α3 is the weight of the interaction term, which is used to capture the interaction between linear and nonlinear outputs.

[0103] Example 2, reference Figure 2 , which is an embodiment of the present invention, provides a system for a safety risk assessment method for a power generation enterprise, including: a data acquisition and preprocessing module, a feature extraction and optimization module, a linear and nonlinear modeling module, a combination and integration module, and a risk assessment and decision support module.

[0104] The data acquisition and preprocessing module collects data in real time from various equipment and environmental sensors of power generation companies, and cleans, normalizes and denoises the collected raw data.

[0105] The feature extraction and optimization module extracts important features from the preprocessed data, optimizes the extracted features by using importance analysis and correlation screening, removes redundant features, and generates a comprehensive feature set.

[0106] The linear and nonlinear modeling module uses a linear regression model to model the feature set, captures the linear relationship in the data and the linear interaction between features, and uses a neural network model and a nonlinear model to process the nonlinear relationship in the data.

[0107] The combination and integration module integrates the output results of linear and nonlinear models, uses nonlinear combination functions to dynamically adjust the impact of the two models, and automatically adjusts the weights and interaction terms in the combination function through training to optimize the final risk assessment output.

[0108] The risk assessment and decision support module determines the current safety status of the system based on the final output of the combination and integration module. The predicted value is compared with the preset threshold to evaluate whether it is within the safety range.

[0109] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., which can store program code.

[0110] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0111] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or, if necessary, processing in another suitable manner, and then stored in a computer memory.

[0112] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, a plurality of steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0113] Embodiment 3: In this embodiment, in order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments. This embodiment experiments the existing traditional method and the method of this embodiment respectively.

[0114] The main purpose of this simulation test is to verify the effectiveness of the safety risk assessment system for power generation enterprises described in the present invention. Specifically, by comparing with traditional safety risk assessment methods, the accuracy of risk prediction, real-time performance and robustness of the model of the system of the present invention under different power generation equipment operating environments are evaluated. The test will simulate the actual operation of power generation enterprises, and demonstrate the advantages of the system of the present invention in complex environments through the entire process of data collection, feature extraction and optimization, model construction and integration, and risk assessment and decision support.

[0115] The test will be carried out in a large thermal power plant equipped with multiple important equipment, including steam turbines, generator sets, cooling systems, and transmission equipment. The test will cover different operating loads and environmental conditions to fully evaluate the performance of the system under variable conditions. Specifically, the test will be carried out under three different load conditions, namely light load (30% load), medium load (60% load) and full load (100% load). In addition, the test will consider two extreme environmental conditions: high temperature and high humidity environment in summer (temperature 40°C, humidity 80%) and low temperature and low humidity environment in winter (temperature -10°C, humidity 30%). These environmental conditions and load combinations simulate the most challenging operating conditions that power generation companies may encounter, ensuring the comprehensiveness and reliability of the test results.

[0116] During the test, a wide range of equipment operation data and environmental data will be collected, including voltage, current, power, speed, temperature, vibration, load, cooling water temperature, ambient temperature, humidity, wind speed, etc. These data will be analyzed through the multi-level processing and dynamic combination model of the system of the present invention to generate safety risk assessment results.

[0117] Table 1 Data comparison table

[0118]

[0119]

[0120] The test results show that under various operating conditions, the system of the present invention is significantly better than the traditional linear regression model and simple statistical model, especially in the high temperature and high humidity environment in summer. The traditional method fails to capture the complex nonlinear relationship, resulting in generally large risk prediction errors. Under light load, full load and medium load operation, the errors of the traditional linear regression model and statistical model range from 16.67% to 28.57%, while the error of the system of the present invention is controlled within 10.00%, with the lowest being only 5.56%. This shows that through multi-level feature extraction and dynamic combination, the system of the present invention can more comprehensively analyze the complex relationships in the data and perform more accurate risk assessment of the actual operating status of the equipment. Especially under complex environmental conditions, the system of the present invention shows strong robustness and adaptability, providing more accurate and reliable safety guarantees for the equipment operation of power generation enterprises.

[0121] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A safety risk assessment method for power generation enterprises, characterized in that: include: Use sensors to collect power generation enterprise equipment data and environmental data and perform data preprocessing; The preprocessed data is analyzed in batches and layers to extract multi-dimensional features, and a comprehensive feature set is formed through feature optimization; Use linear and nonlinear models to process the comprehensive feature set and output the prediction results; Dynamically combine linear and nonlinear prediction results to output the final risk assessment results.

2. A safety risk assessment method for power generation enterprises according to claim 1, characterized in that: The power generation enterprise equipment data and environmental data include voltage data, current data, power data, speed data, temperature rise data, winding temperature data, load data, cooling water temperature data and cooling water pressure data, external environment temperature data, humidity data, wind speed data and air pressure data; The data preprocessing is to detect missing values ​​on the collected raw data, fill in the missing values ​​with interpolation, remove noise in the data with the Kalman filter method, calculate the standard score of the data with the Z-score method, mark the data points exceeding the set threshold as abnormal data, detect and process the abnormal values ​​with the IQR method, and eliminate the abnormal data; normalize the data of different dimensions, convert the data to a unified numerical range, and classify and batch process the normalized data according to the equipment category and time period.

3. A safety risk assessment method for power generation enterprises according to claim 2, characterized in that: The comprehensive feature set is to process the pre-processed data in batches; Process the equipment data of different categories in batches according to the equipment categories, divide the data according to time periods, extract basic statistical features from the batch-processed data, perform dynamic feature extraction based on the basic statistical features, calculate the change slope, trend direction and fluctuation amplitude of the data, and identify dynamic change features; Advanced feature extraction is performed based on the results of dynamic feature extraction, and the periodic features, modal features and nonlinear relationship features of the data are extracted through nonlinear analysis methods; Optimize based on the results of advanced feature extraction, remove redundant features through importance analysis and correlation screening, and generate a comprehensive feature set; The extracting of basic statistical features includes calculating a weighted mean, a standard deviation, a maximum value, a minimum value and a corrected variance.

4. A safety risk assessment method for power generation enterprises according to claim 3, characterized in that: The dynamic feature extraction calculates the change slope, trend direction and fluctuation amplitude of the data and identifies the dynamic change characteristics. The expression is: Amplitude t =ΔX·(1+skew(X)) Among them, β t is the slope of change of trend adjustment, t i is the i-th time point, is the mean value at a time point, is the adjustment coefficient of the standard deviation, Amplitude e t is the fluctuation range, ΔX is the extreme value difference, skew(X) is the skewness of the data, AR(X) is the autoregressive model, φ0, φ k is the autoregressive coefficient, the weight in the autoregressive model, X t-k is the data value lagged k periods later, ε t is the noise term, Trend AR It is the direction of autoregressive trend. When φ1>0, the current data is positively correlated with the data at the previous time point. When φ1<0, the current data is negatively correlated with the data at the previous time point. When φ1=0, there is no correlation between the current data and the data at the previous time point. The periodic characteristics, modal characteristics and nonlinear relationship characteristics of the data are extracted by nonlinear analysis method, and the expression is: Where P(f) is the energy spectral density of the periodic characteristic, λ is the attenuation factor, and f k is the kth frequency component, f k-1 is the k-1th frequency component, β t is the slope of change after trend adjustment, M(t) is the modal characteristic, a n is the amplitude of the nth mode, f n is the frequency of the nth mode, φ n is the phase of the nth mode, g(t) is the attenuation function, Trend AR is the trend direction, K(X,Y) is the dynamic kernel function, which represents the nonlinear relationship between the two data sets, and α i , α j is the characteristic coefficient, ||X i -Y j || 2 is the Euclidean distance, which represents the distance between two data points, σ ​​is the kernel width, which controls the smoothness of the kernel function, and Amplitude t is the fluctuation range; The redundant features are removed through importance analysis and correlation screening to generate a comprehensive feature set, which is expressed as: FeatureSet opt ={f select ∣Importance(f i )>θ} Among them, Importance(f i ) is the contribution of the feature to the result, Var adj (f i ) is the corrected variance of the i-th feature, ∑Var adj (f j ) is the sum of the corrected variances of all features, β t is the slope of change after trend adjustment, ρ ij The feature f i and f j The correlation coefficient between the two features indicates the linear relationship between the two features. Cov(f i ,f j ) is the feature f i and f j The covariance between The features f i and f j The standard deviation of δ is the correlation threshold, f select FeatureSet is the filtered feature set. opt is a comprehensive feature set, θ is the importance threshold, and important features are screened.

5. A safety risk assessment method for power generation enterprises according to claim 4, characterized in that: The use of linear and nonlinear models to process the comprehensive feature set respectively includes using the least squares method to find the intercept in the linear regression model, and then processing the comprehensive feature set through the linear regression model to generate a linear prediction output, which is expressed as: Among them, β0 is the intercept of the model, β i is the i-th feature f i The regression coefficient, f i is the i-th feature in the comprehensive feature set, γ j is the coefficient of the interaction term between features in the linear function, indicating that two features f i and f j The impact of the interaction between linear is the error term of the linear model; The neural network model is used to process the comprehensive feature set to generate nonlinear prediction output, which is expressed as: y nonlinear =NN(F) Among them, NN(F) is the neural network model, and F is the overall input of the high-level optimization feature set.

6. A safety risk assessment method for power generation enterprises according to claim 5, characterized in that: The dynamic combination of linear and nonlinear prediction results is to dynamically combine the linear prediction output and the nonlinear prediction output through a nonlinear integration function, and introduce the interaction terms between the features. After training, the weights and interaction terms in the combination function are dynamically adjusted. The expression is: Among them, y final is the final prediction output, g(·) is the nonlinear integration function, h(y linear ,y nonlinear ) is the combination function, λ ij is the coefficient of the feature interaction term in the nonlinear function, f i ·f j The feature f i and f j The interaction terms between .

7. A safety risk assessment method for power generation enterprises according to claim 6, characterized in that: The risk assessment result is obtained by comparing the result of the dynamic combination output of the linear prediction output and the nonlinear prediction output with the threshold. If the result of the dynamic combination output is greater than the threshold, it is determined to be a dangerous state. If the result of the dynamic combination output is less than or equal to the threshold, it is determined to be a safe state.

8. A system using a safety risk assessment method for a power generation enterprise as claimed in any one of claims 1 to 7, characterized in that: It includes data acquisition and preprocessing module, feature extraction and optimization module, linear and nonlinear modeling module, combination and integration module, and risk assessment and decision support module; The data acquisition and preprocessing module collects data from various equipment and environmental sensors of power generation enterprises in real time, and cleans, normalizes and denoises the collected raw data; The feature extraction and optimization module extracts important features from the preprocessed data, optimizes the extracted features by using importance analysis and correlation screening, removes redundant features, and generates a comprehensive feature set; The linear and nonlinear modeling module uses a linear regression model to model the feature set, captures the linear relationship in the data and the linear interaction between the features, and uses a neural network model and a nonlinear model to process the nonlinear relationship in the data; The combination and integration module integrates the output results of the linear and nonlinear models, uses a nonlinear combination function to dynamically adjust the impact of the two models, and automatically adjusts the weights and interaction terms in the combination function through training to optimize the final risk assessment output; The risk assessment and decision support module determines the current safety status of the system based on the final output of the combination and integration module, compares the predicted value with the preset threshold, and evaluates whether it is within the safety range.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of a safety risk assessment method for a power generation enterprise described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of a safety risk assessment method for a power generation enterprise as described in any one of claims 1 to 7 are implemented.

Images