Boiler health assessment system and method based on data mining technology
The boiler health assessment system, which utilizes data mining technology and combines entropy weight method with improved density peak clustering and isolated forest algorithms, achieves real-time assessment and anomaly detection of boiler health status. This solves the problems of untimely assessment and misjudgment in traditional methods and improves equipment management efficiency.
Patent Information
- Application Number
- CN202611266053.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-08-20
- Publication Date
- 2026-11-06
AI Technical Summary
Existing technologies are insufficient for effectively and timely assessing the health status of boilers, resulting in untimely equipment maintenance and troubleshooting. Furthermore, traditional methods are limited by individual capabilities and cannot cope with complex operating environments and equipment aging issues.
A boiler health assessment system based on data mining technology is adopted, including modules for data acquisition, preprocessing, feature dimensionality reduction, pattern recognition, and report generation. It combines entropy weight method, principal component analysis, improved density peak clustering, and isolated forest algorithm to achieve real-time assessment and anomaly detection.
It provides efficient, real-time boiler health assessment, reduces edge detection errors and missed detections, ensures safe and efficient equipment operation, and supports multi-platform data synchronization and personalized report generation.
Smart Images
Figure CN123049414A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of boiler technology, and more specifically, to a boiler health assessment system and method based on data mining technology. Background Technology
[0002] As a heat energy supply device, the boiler occupies a very important position in industrial production. In order to operate the boiler safely, efficiently and environmentally, it is necessary to carry out regular or even real-time testing and evaluation.
[0003] The complexity of this boiler equipment, the variability of its operating conditions, and the limitations of its testing system make evaluation difficult. Traditionally, boiler economics and operational health are judged by unit consumption and reverse balance thermal efficiency. However, due to changes in the external operating environment and equipment aging, the unit consumption curve is dynamic, requiring periodic testing, which is time-consuming and often disrupts normal production. Furthermore, addressing abnormal parameter changes, equipment fault diagnosis, and emergency handling during equipment operation is hampered by limitations in individual capabilities, leading to delays in equipment maintenance, defect and fault handling. Therefore, an effective method for analyzing and evaluating the healthy operation of boiler equipment is needed. Summary of the Invention
[0004] The purpose of this invention is to provide a boiler health assessment system and method based on data mining technology to solve the problems mentioned in the background art.
[0005] To address the aforementioned technical problems, one objective of this invention is to provide a boiler health assessment system based on data mining technology, comprising:
[0006] The data acquisition module is used to collect boiler operating status parameters in real time, including temperature, pressure, flow rate, fuel consumption, power consumption and water consumption.
[0007] The data preprocessing module is used to perform missing value imputation, noise filtering, and normalization on the collected data;
[0008] The data extraction module is used to perform feature dimensionality reduction on the preprocessed data by combining entropy weighting and principal component analysis.
[0009] The data mining module is used to combine improved density peak clustering and isolated forest to identify boiler operating patterns and potential anomalies;
[0010] The assessment management module is used to publish assessment models and create assessment models applicable to different boiler product models.
[0011] The model selection module is used to match and associate the evaluation model with the boiler product model;
[0012] The report generation module is used to automatically generate health assessment reports that include energy conservation and environmental protection assessments as well as safe operation assessments.
[0013] The communication module is used to transmit boiler operation data to the cloud platform in real time and receive remote control commands.
[0014] As a further improvement to this technical solution, the data preprocessing module uses sliding window mean interpolation to remove missing values, and the specific steps for eliminating noise using wavelet thresholding are as follows:
[0015] A1. Select the wavelet basis function, using the Mexican Hat wavelet, and calculate the formula as follows:
[0016]
[0017] in, Represents the Mexican Hat wavelet function. It is a time variable;
[0018] A2. Determine the scale range for the analysis; the calculation formula is as follows:
[0019]
[0020]
[0021]
[0022] in, The actual frequency of the signal. The center frequency of the wavelet basis function. The sampling frequency of the signal. For scale parameters, For the largest scale, The minimum frequency set. For the smallest scale, The set maximum frequency;
[0023] A3. Generate a family of wavelets and calculate the wavelet coefficients. The calculation formula is as follows:
[0024]
[0025]
[0026] in, Indicates wavelet family, Indicates the translation parameter. Indicated in scale parameter Translation parameters wavelet coefficients at the location, Indicates the input signal;
[0027] A4. Noise removal using a threshold is calculated using the following formula:
[0028]
[0029]
[0030] in, Indicates the threshold. express function, For adjustment coefficients, For general thresholds, The standard deviation of noise. This is the signal length.
[0031] As a further improvement to this technical solution, the calculation formula for normalization processing in the data preprocessing module is as follows:
[0032]
[0033]
[0034]
[0035] in, For the first The first sample 1 eigenvalue, Represents the total sample size. for The normalized value, For the first The mean of each feature, For the first The standard deviation of each feature.
[0036] As a further improvement to this technical solution, the specific steps for feature dimensionality reduction processing of the preprocessed data in the data extraction module are as follows:
[0037] B1. Calculate the information entropy of each indicator. The calculation formula is as follows:
[0038]
[0039]
[0040] in, Indicates the first The first indicator in the The proportion in each sample Indicates the smoothing term. Indicates the first Information entropy of each indicator;
[0041] B2. Calculate the difference coefficient and weight. The calculation formula is as follows:
[0042]
[0043]
[0044] in, Indicates the first The coefficient of difference for each indicator Indicates the first The weight of each indicator, Indicates the number of indicators;
[0045] B3. The effect of high-information-content indicators is amplified through weighting; the calculation formula is as follows:
[0046]
[0047] in, Indicates the weighted number of... The first sample One eigenvalue;
[0048] B4. Calculate the covariance matrix, perform eigenvalue decomposition on the covariance matrix, select principal components, and project the data.
[0049] As a further improvement to this technical solution, the specific steps of the data mining module in identifying boiler operating modes and potential anomalies are as follows:
[0050] C1. Identifying boiler operation modes based on improved density peak clustering;
[0051] C2. Identify potential abnormal operating states based on isolated forests.
[0052] As a further improvement to this technical solution, the specific steps of C1 are as follows:
[0053] C1.1 Calculate the distance between all sample points using the Euclidean distance metric to obtain the distance matrix. ;
[0054] C1.2 Determine the cutoff distance Choose a neighborhood that contains an average of 1% of the total data for each point;
[0055] C1.3 Calculate the local density using the following formula:
[0056]
[0057] in, Indicates the first Local density of each sample point Represents sample points and sample points The Euclidean distance between them;
[0058] C1.4 Improve relative distance calculation by introducing density difference weights to avoid misjudging edge samples. For each sample point, calculate its weighted minimum distance to higher density points using the following formula:
[0059]
[0060]
[0061] in, Indicates the first Local density of each sample point This indicates taking the minimum value. This indicates taking the maximum value. Indicates the first The relative distance between sample points Indicates the decision value;
[0062] C1.5. In the decision diagram, select... and Sample points that are significantly higher than their surrounding points are used as cluster centers. For non-center points, they are assigned to the nearest and densest cluster center.
[0063] As a further improvement to this technical solution, the specific steps of C2 are as follows:
[0064] C2.1 Input the dataset after C1 clustering, along with cluster labels;
[0065] C2.2. Construct an isolated forest model for each cluster separately, adjust the subsampling amount based on the local density of the cluster centers, and perform subsampling operations. The calculation formula is as follows:
[0066]
[0067] in, Indicates the first Subsampling quantity corresponding to each cluster center Indicates the basic subsample size. Indicates the first Local density of cluster centers Represents the maximum local density. Indicates the regulating factor;
[0068] C2.3 Set the number of trees, randomly select features, and randomly select split points within the feature value range;
[0069] C2.4 Calculate the path length in each tree, introduce density weights, and adjust the path length. The calculation formula is as follows:
[0070]
[0071]
[0072] in, This indicates the corrected path length. Indicates path length. Indicates density weight;
[0073] C2.5 Calculate the outlier score. Convert the corrected path length into an outlier score of [0,1]. If the outlier score is close to 1, it is judged as an outlier point. If the outlier score is close to 0, it is judged as a normal point.
[0074] As a further improvement to this technical solution, the specific steps for the model selection module to match and associate the evaluation model with the boiler product model are as follows:
[0075] D1. Establish association rules through the matching engine of the product model library and the evaluation model library;
[0076] D2. After model matching is completed, verify the sensor coverage of the data acquisition module;
[0077] D3. Monitor boiler operating data fluctuations in real time during the evaluation period.
[0078] As a further improvement to this technical solution, the report generation module includes:
[0079] Multi-dimensional search unit: Supports searching evaluation reports by customer geographic distribution, product model, and equipment number, and generates visual charts, including health score radar charts, anomaly distribution heat maps, and historical trend comparison curves;
[0080] Custom assessment unit: Allows users to select one or more boilers across devices, set any assessment period, and trigger the instant generation of health assessment reports;
[0081] Compliance Verification Unit: Automatically references standards to verify the compliance of the model and marks the verification results and referenced clauses in the report;
[0082] Multi-terminal synchronization unit: The evaluation report can be exported to a unified format with one click and pushed synchronously to PC, mobile APP and smart terminal through the communication module to ensure data consistency across multiple platforms.
[0083] The second objective of this invention is to provide a boiler health assessment method based on data mining technology. The boiler health assessment system based on the aforementioned data mining technology includes the following steps:
[0084] S1. Real-time acquisition of boiler operating status parameters, including temperature, pressure, flow rate, fuel consumption, power consumption and water consumption;
[0085] S2. Perform missing value imputation, noise filtering, and normalization on the collected data;
[0086] S3. Combining entropy weighting and principal component analysis, feature dimensionality reduction is performed on the preprocessed data.
[0087] S4. Identify boiler operation modes and potential anomalies based on improved density peak clustering and isolated forest;
[0088] S5. Publish the evaluation model and create an evaluation model applicable to different boiler product models;
[0089] S6. Match and associate the evaluation model with the boiler product model;
[0090] S7. Automatically generate a health assessment report that includes energy-saving and environmental protection assessments and safe operation assessments;
[0091] S8 transmits boiler operation data to the cloud platform in real time and receives remote control commands.
[0092] Compared with existing technologies, the beneficial effects of this invention are as follows: In this boiler health assessment system and method based on data mining technology, sliding window mean interpolation and wavelet threshold denoising techniques are used to effectively eliminate noise interference while repairing missing data, and normalization processing is used to eliminate dimensional differences, laying a high-quality data foundation for subsequent analysis; the data extraction module combines information entropy weighting and principal component analysis to focus on high-value features, and the data mining module integrates improved density peak clustering and density-weighted isolated forest algorithms, optimizes the selection of cluster centers through local density difference weighting, and adaptively adjusts the anomaly detection strategy for different operating modes, effectively solving the problems of edge misjudgment and missed detection in traditional methods under complex operating conditions. Attached Figure Description
[0093] Figure 1 This is a system structure diagram of the present invention;
[0094] Figure 2 This is a structural diagram of the report generation module in this invention;
[0095] Figure 3 This is a flowchart of the method of the present invention;
[0096] The meanings of the various markers in the diagram are as follows: 10, Data Acquisition Module; 20, Data Preprocessing Module; 30, Data Extraction Module; 40, Data Mining Module; 50, Evaluation Management Module; 60, Model Selection Module; 70, Report Generation Module; 700, Multi-Dimensional Retrieval Unit; 701, Custom Evaluation Unit; 702, Compliance Verification Unit; 703, Multi-Terminal Synchronization Unit; 80, Communication Module. Detailed Implementation
[0097] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.
[0098] like Figure 1 and Figure 2 As shown, this embodiment provides a boiler health assessment system based on data mining technology, including:
[0099] Data acquisition module 10: used to collect boiler operating status parameters in real time, including temperature, pressure, flow rate, fuel consumption, power consumption and water consumption;
[0100] Data preprocessing module 20: used to perform missing value imputation, noise filtering and normalization on the collected data;
[0101] Data extraction module 30: used to perform feature dimensionality reduction on preprocessed data by combining entropy weight method and principal component analysis;
[0102] Data mining module 40: used to combine improved density peak clustering and isolated forest to identify boiler operating patterns and potential anomalies;
[0103] Evaluation Management Module 50: Used to publish evaluation models and create evaluation models applicable to different boiler product models;
[0104] Model matching module 60: Used to match and associate the evaluation model with the boiler product model;
[0105] Report generation module 70: Used to automatically generate health assessment reports that include energy conservation and environmental protection assessments and safe operation assessments;
[0106] Communication module 80: Used to transmit boiler operation data to the cloud platform in real time and receive remote control commands.
[0107] In this embodiment, the data preprocessing module 20 uses sliding window mean interpolation to remove missing values, and the specific steps for using wavelet thresholding to eliminate noise are as follows:
[0108] A1. Select the wavelet basis function, using the Mexican Hat wavelet. The Mexican Hat wavelet is the second derivative of the Gaussian function, and its waveform exhibits a central peak and rapidly decaying side lobes. This shape highly matches the pulse noise commonly found in boiler sensor signals. The calculation formula is as follows:
[0109]
[0110] in, The Mexican Hat wavelet function, with its symmetry and rapid decay characteristics, can accurately locate instantaneous abrupt changes in signals. In boiler multi-sensor systems, the noise characteristics of different parameters vary significantly. The versatility of the Mexican Hat wavelet reduces the need for dedicated filters for different sensors, simplifying engineering deployment. It is a time variable;
[0111] A2. Determine the scale range of the analysis to cover the full frequency band characteristics of boiler operation data and avoid the loss of key frequency components. The calculation formula is as follows:
[0112]
[0113]
[0114]
[0115] in, The actual frequency of the signal. The center frequency of the wavelet basis function. The sampling frequency of the signal. For scale parameters, To capture the boiler's long-cycle operation mode at the largest possible scale, The minimum frequency set. To capture transient events such as valve start-up and shutdown, and combustion oscillations at the smallest scale. The set maximum frequency;
[0116] A3. Generate wavelet families and calculate wavelet coefficients to decompose the multi-scale features of the signal, providing a fine-grained analytical basis for noise separation. The calculation formula is as follows:
[0117]
[0118]
[0119] in, Indicates wavelet family, Indicates the translation parameter. Indicated in scale parameter Translation parameters wavelet coefficients at the location, The input signal is represented by wavelet coefficients at different scales, which reflect the global trend and local details of the boiler signal, respectively. High-frequency noise has a significant amplitude in small-scale coefficients, while the real signal dominates in large-scale coefficients.
[0120] A4. Noise removal through thresholding achieves precise noise elimination while preserving the true signal. The calculation formula is as follows:
[0121]
[0122]
[0123] in, Indicates the threshold. express function, For adjustment coefficients, For general thresholds, The standard deviation of noise. Given the signal length, and combined with an improved soft threshold function, wavelet coefficients below the threshold are zeroed out, directly filtering out random noise.
[0124] In this embodiment, the calculation formula for normalization processing in the data preprocessing module 20 is as follows:
[0125]
[0126]
[0127]
[0128] in, For the first The first sample 1 eigenvalue, Represents the total sample size. for The normalized value, For the first The mean of each feature, For the first The standard deviation of each feature.
[0129] In this embodiment, the specific steps of feature dimensionality reduction processing of the preprocessed data in the data extraction module 30 are as follows:
[0130] B1. Calculate the information entropy of each indicator to quantify the information content of the features and screen key indicators sensitive to boiler status. The lower the entropy value, the more uneven the feature distribution and the higher the information content; the higher the entropy value, the more uniform the feature distribution and the lower the information content. The calculation formula is:
[0131]
[0132]
[0133] in, Indicates the first The first indicator in the The proportion in each sample Indicates the smoothing term. Indicates the first Information entropy of each indicator;
[0134] B2. Calculate the difference coefficient and weights, assigning higher weights to features with high information content to enhance their contribution to subsequent analysis. The calculation formula is as follows:
[0135]
[0136]
[0137] in, Indicates the first The coefficient of difference for each indicator Indicates the first The weight of each indicator, Indicates the number of indicators;
[0138] B3. By amplifying the role of high-information-content indicators through weighting, the projection direction of principal component analysis is optimized. Traditional PCA may deviate from the true principal components due to features with similar variances. After weighting, the variance of high-weight features is amplified, dominating the direction of the principal components. The calculation formula is as follows:
[0139]
[0140] in, Indicates the weighted number of... The first sample One eigenvalue;
[0141] B4. Calculate the covariance matrix, perform eigenvalue decomposition on the covariance matrix, select principal components, project the data, and compress the data dimensionality while retaining key information.
[0142] In this embodiment, the specific steps of the data mining module 40 in identifying boiler operating modes and potential anomalies are as follows:
[0143] C1. Based on improved density peak clustering to identify boiler operation modes, boiler operation data is divided into several typical operating mode to provide a benchmark reference for subsequent anomaly detection;
[0144] C1.1. Use Euclidean distance metric to calculate the distance between all sample points, quantify the similarity between boiler operating states, and obtain the distance matrix. ;
[0145] C1.2 Determine the cutoff distance The neighborhood range for density calculation is adaptively set to balance clustering sensitivity and noise resistance, and the neighborhood of each point is selected to contain an average of 1% of the total data.
[0146] C1.3 Calculate local density, identify dense areas of boiler operation, and locate the center of typical operating conditions. The calculation formula is as follows:
[0147]
[0148] in, Indicates the first Local density of each sample point Represents sample points and sample points The Euclidean distance between them;
[0149] C1.4 Improves relative distance calculation by introducing density difference weights to avoid misclassification of edge samples, solving the problem of misclassification of edge samples in traditional methods, and improving the clarity of cluster boundaries. For each sample point, calculate its weighted minimum distance to higher density points using the following formula:
[0150]
[0151]
[0152] in, Indicates the first Local density of each sample point This indicates taking the minimum value. This indicates taking the maximum value. Indicates the first The relative distance between sample points Indicates the decision value;
[0153] C1.5. In the decision diagram, select... and Sample points that are significantly higher than their surrounding points are used as cluster centers. For non-center points, they are assigned to the nearest and densest cluster centers to complete the division of operating modes and provide a benchmark label for anomaly detection.
[0154] C2. Based on the isolation forest, find potential abnormal operating states, and adaptively detect abnormal points that deviate from the normal mode under the clustering division of the working conditions.
[0155] C2.1 Input the dataset after C1 clustering, and attach cluster labels to realize anomaly detection under different working conditions, so as to avoid insufficient sensitivity of global detection;
[0156] C2.2. Construct an isolated forest model for each cluster, adjust the subsampling amount based on the local density of the cluster centers, and perform subsampling operations. Adjust the sample size according to the working condition density to improve the detection sensitivity of sparse working conditions. The calculation formula is as follows:
[0157]
[0158] in, Indicates the first Subsampling quantity corresponding to each cluster center Indicates the basic subsample size. Indicates the first Local density of cluster centers Represents the maximum local density. Indicates the regulating factor;
[0159] C2.3 Set the number of trees, randomly select features, and randomly select split points within the feature value range. Generate path length differences through random splitting to quantify the degree of sample anomalies.
[0160] C2.4 Calculate the path length in each tree, introduce density weights, correct the path length, suppress false alarms in high-density areas, and enhance the detection sensitivity in low-density areas. The calculation formula is as follows:
[0161]
[0162]
[0163] in, This indicates the corrected path length. Indicates path length. Indicates density weight;
[0164] C2.5 Calculate the outlier score. Convert the corrected path length into an outlier score of [0,1]. If the outlier score is close to 1, it is judged as an outlier point. If the outlier score is close to 0, it is judged as a normal point.
[0165] In this embodiment, the specific steps of the model selection module 60 in matching and associating the evaluation model with the boiler product model are as follows:
[0166] D1. By using the matching engine of the product model library and the evaluation model library, association rules are established to realize the automated matching of boiler product models and evaluation models, thus solving the management complexity of multiple models of equipment. The model library stores boiler model metadata, the model library pre-trains model parameters, and the feature mapping in the rule engine matches the model through key parameters. When multiple models are matched, the model with the most training data covering the model is selected.
[0167] D2. After the model matching is completed, verify the sensor coverage of the data acquisition module 10, extract the list of input features required by the model, and remind the user when key sensors are missing.
[0168] D3. During the evaluation period, monitor boiler operating data fluctuations in real time, dynamically track changes in the statistical characteristics of boiler operating data, and trigger model refitting to cope with operating condition drift, including fluctuation index calculation and abnormal fluctuation detection.
[0169] In this embodiment, the report generation module 70 includes:
[0170] Multi-dimensional search unit 700: Supports searching evaluation reports by customer geographic distribution, product model, and equipment number, and generates visual charts, including health score radar charts, anomaly distribution heat maps, and historical trend comparison curves;
[0171] Customized Assessment Unit 701: Allows users to select one or more boilers across devices, set any assessment period, trigger the instant generation of health assessment reports, realize the rapid location and visualization of boiler health assessment reports, support multi-perspective data insights, provide flexible report generation strategies, and meet personalized analysis needs.
[0172] Compliance Verification Unit 702: Automatically references standards to verify the compliance of the model and marks the verification results and referenced clauses in the report to ensure that the assessment results meet industry standards and regulatory requirements and avoid compliance risks;
[0173] Multi-terminal synchronization unit 703: Exports the assessment report to a unified format with one click and pushes it to PC, mobile APP and smart terminal through communication module 80, ensuring data consistency across multiple platforms, realizing seamless push and format adaptation of the assessment report across all platforms, and improving information accessibility.
[0174] like Figure 3 As shown, this embodiment also provides a boiler health assessment method based on data mining technology. Based on the aforementioned boiler health assessment system based on data mining technology, the method includes the following steps:
[0175] S1. Real-time acquisition of boiler operating status parameters, including temperature, pressure, flow rate, fuel consumption, power consumption and water consumption;
[0176] S2. Perform missing value imputation, noise filtering, and normalization on the collected data;
[0177] S3. Combining entropy weighting and principal component analysis, feature dimensionality reduction is performed on the preprocessed data.
[0178] S4. Identify boiler operation modes and potential anomalies based on improved density peak clustering and isolated forest;
[0179] S5. Publish the evaluation model and create an evaluation model applicable to different boiler product models;
[0180] S6. Match and associate the evaluation model with the boiler product model;
[0181] S7. Automatically generate a health assessment report that includes energy-saving and environmental protection assessments and safe operation assessments;
[0182] S8 transmits boiler operation data to the cloud platform in real time and receives remote control commands.
[0183] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely preferred examples and are not intended to limit the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.
Claims
1. A boiler health assessment system based on data mining technology, characterized in that, include: The data acquisition module (10) is used to collect boiler operating status parameters in real time, including temperature, pressure, flow rate, fuel consumption, power consumption and water consumption; The data preprocessing module (20) is used to perform missing value imputation, noise filtering and normalization on the collected data; The data extraction module (30) is used to perform feature dimensionality reduction on the preprocessed data by combining entropy weight method and principal component analysis; The data mining module (40) is used to combine improved density peak clustering and isolated forest to identify boiler operating patterns and potential anomalies; The assessment management module (50) is used to publish assessment models and create assessment models applicable to different boiler product models; The model selection module (60) is used to match and associate the evaluation model with the boiler product model; The report generation module (70) is used to automatically generate a health assessment report that includes energy conservation and environmental protection assessments and safe operation assessments; The communication module (80) is used to transmit boiler operation data to the cloud platform in real time and receive remote control commands.
2. The boiler health assessment system based on data mining technology according to claim 1, characterized in that, The data preprocessing module (20) uses sliding window mean interpolation to remove missing values and wavelet thresholding to eliminate noise. The specific steps are as follows: A1. Select the wavelet basis function, using the Mexican Hat wavelet, and calculate the formula as follows: in, Represents the Mexican Hat wavelet function. It is a time variable; A2. Determine the scale range for the analysis; the calculation formula is as follows: in, The actual frequency of the signal. The center frequency of the wavelet basis function. The sampling frequency of the signal. For scale parameters, For the largest scale, The minimum frequency set. For the smallest scale, The set maximum frequency; A3. Generate a family of wavelets and calculate the wavelet coefficients. The calculation formula is as follows: in, Indicates wavelet family, Indicates the translation parameter. Indicated in scale parameter Translation parameters wavelet coefficients at the location, Indicates the input signal; A4. Noise removal using a threshold is calculated using the following formula: in, Indicates the threshold. express function, For adjustment coefficients, For general thresholds, The standard deviation of noise. This is the signal length.
3. The boiler health assessment system based on data mining technology according to claim 2, characterized in that, The calculation formula for normalization processing in the data preprocessing module (20) is as follows: in, For the first The first sample 1 eigenvalue, Represents the total sample size. for The normalized value, For the first The mean of each feature, For the first The standard deviation of each feature.
4. The boiler health assessment system based on data mining technology according to claim 1, characterized in that, The specific steps for feature dimensionality reduction processing of the preprocessed data in the data extraction module (30) are as follows: B1. Calculate the information entropy of each indicator. The calculation formula is as follows: in, Indicates the first The first indicator in the The proportion in each sample Indicates the smoothing term. Indicates the first Information entropy of each indicator; B2. Calculate the difference coefficient and weight. The calculation formula is as follows: in, Indicates the first The coefficient of difference for each indicator Indicates the first The weight of each indicator, Indicates the number of indicators; B3. The effect of high-information-content indicators is amplified through weighting; the calculation formula is as follows: in, Indicates the weighted number of... The first sample One eigenvalue; B4. Calculate the covariance matrix, perform eigenvalue decomposition on the covariance matrix, select principal components, and project the data.
5. The boiler health assessment system based on data mining technology according to claim 1, characterized in that, The specific steps of the data mining module (40) in identifying boiler operating modes and potential anomalies are as follows: C1. Identifying boiler operation modes based on improved density peak clustering; C2. Identify potential abnormal operating states based on isolated forests.
6. The boiler health assessment system based on data mining technology according to claim 5, characterized in that, The specific steps of C1 are as follows: C1.1 Calculate the distance between all sample points using the Euclidean distance metric to obtain the distance matrix. ; C1.2 Determine the cutoff distance Choose a neighborhood that contains an average of 1% of the total data for each point; C1.3 Calculate the local density using the following formula: in, Indicates the first Local density of each sample point Represents sample points and sample points The Euclidean distance between them; C1.4 Improve relative distance calculation by introducing density difference weights to avoid misjudging edge samples. For each sample point, calculate its weighted minimum distance to higher density points using the following formula: in, Indicates the first Local density of each sample point This indicates taking the minimum value. This indicates taking the maximum value. Indicates the first The relative distance between sample points Indicates the decision value; C1.
5. In the decision diagram, select... and Sample points that are significantly higher than their surrounding points are used as cluster centers. For non-center points, they are assigned to the nearest cluster center with higher density.
7. The boiler health assessment system based on data mining technology according to claim 6, characterized in that, The specific steps of C2 are as follows: C2.1 Input the dataset after C1 clustering, along with cluster labels; C2.
2. Construct an isolated forest model for each cluster separately, adjust the subsampling amount based on the local density of the cluster centers, and perform subsampling operations. The calculation formula is as follows: in, Indicates the first The number of subsamples corresponding to each cluster center point Indicates the basic subsample size. Indicates the first Local density of cluster centers Represents the maximum local density. Indicates the regulating factor; C2.3 Set the number of trees, randomly select features, and randomly select split points within the feature value range; C2.4 Calculate the path length in each tree, introduce density weights, and adjust the path length. The calculation formula is as follows: in, This indicates the corrected path length. Indicates path length. Indicates density weight; C2.5 Calculate the outlier score. Convert the corrected path length into an outlier score of [0,1]. If the outlier score is close to 1, it is judged as an outlier point. If the outlier score is close to 0, it is judged as a normal point.
8. The boiler health assessment system based on data mining technology according to claim 1, characterized in that, The specific steps of the model selection module (60) in matching and associating the evaluation model with the boiler product model are as follows: D1. Establish association rules through the matching engine of the product model library and the evaluation model library; D2. After the model matching is completed, verify the sensor coverage of the data acquisition module (10); D3. Monitor boiler operating data fluctuations in real time during the evaluation period.
9. The boiler health assessment system based on data mining technology according to claim 1, characterized in that, The report generation module (70) includes: Multi-dimensional search unit (700): Supports searching evaluation reports by customer geographic distribution, product model, and equipment number, and generates visual charts, including health score radar chart, anomaly distribution heat map and historical trend comparison curve; Custom assessment unit (701): Allows users to select one or more boilers across devices, set any assessment period, and trigger the instant generation of health assessment reports; Compliance Verification Unit (702): Automatically references standards to verify the compliance of the model and marks the verification results and referenced clauses in the report; Multi-terminal synchronization unit (703): Exports the evaluation report in a unified format with one click and pushes it synchronously to PC, mobile APP and smart terminal through the communication module (80) to ensure data consistency across multiple platforms.
10. A boiler health assessment method based on data mining technology, based on the boiler health assessment system based on data mining technology as described in any one of claims 1-9, characterized in that, Includes the following steps: S1. Real-time acquisition of boiler operating status parameters, including temperature, pressure, flow rate, fuel consumption, power consumption and water consumption; S2. Perform missing value imputation, noise filtering, and normalization on the collected data; S3. Combining entropy weighting and principal component analysis, feature dimensionality reduction is performed on the preprocessed data. S4. Identify boiler operation modes and potential anomalies based on improved density peak clustering and isolated forest; S5. Publish the evaluation model and create an evaluation model applicable to different boiler product models; S6. Match and associate the evaluation model with the boiler product model; S7. Automatically generate a health assessment report that includes energy-saving and environmental protection assessments and safe operation assessments; S8 transmits boiler operation data to the cloud platform in real time and receives remote control commands.