Heat exchanger fault diagnosis method and system based on data dimension reduction

Through data dimensionality reduction technology, the extraction of heat exchanger fault diagnosis method solves the problem of high-dimensional data redundancy and information overlap, improves the accuracy and efficiency of fault diagnosis, and achieves more reliable heat exchanger fault identification.

CN119004282AInactive Publication Date: 2024-11-22WUHAN UNIV OF TECH +1

Patent Information

Application Number
CN202411175640.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-26
Publication Date
2024-11-22
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

Existing heat exchanger fault diagnosis methods rely on manual inspection and empirical judgment, and high-dimensional data leads to redundancy and information overlap, reducing diagnostic accuracy and efficiency.

Method used

采用基于数据降维的方法,通过特征提取和堆叠降噪自动编码器(SDAE)对数据进行降维,提高特征独立性,并使用随机森林模型进行故障诊断。

Benefits of technology

Effectively remove data noise and redundancy, improve diagnostic accuracy and efficiency, reduce model training time and computing resources, avoid overfitting, and improve adaptability and generalization ability to different working conditions.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119004282A_ABST
    Figure CN119004282A_ABST
Patent Text Reader

Abstract

The invention discloses a heat exchanger fault diagnosis method and system based on data dimension reduction, and the method comprises the following steps: collecting the operation data of a heat exchanger under various working conditions, and constructing a data set according to the operation data and the working conditions corresponding to the operation data; extracting features in the data set to obtain feature data, performing dimension reduction on the feature data to obtain a training set, and training a classification model by using the training set; the method comprises the following steps: collecting operation data of a heat exchanger to be diagnosed, extracting features of the operation data of the heat exchanger to be diagnosed, carrying out dimension reduction on the features to obtain dimension-reduced data, inputting the dimension-reduced data into a trained classification model, and outputting a state and a working condition of the heat exchanger to be diagnosed by the classification model, fault diagnosis of the heat exchanger is realized. According to the method, the manpower demand in industrial production can be reduced, the maintenance time after the fault occurs is shortened, and the fault diagnosis accuracy is improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of heat exchanger fault diagnosis, and in particular to a heat exchanger fault diagnosis method and system based on data dimension reduction. Background Art

[0002] As a key heat transfer device in modern industry, heat exchangers are widely used in many industries such as chemical industry, power generation, food processing, shipbuilding and air conditioning. Due to the special structure and complex working environment, heat exchangers are prone to internal tube scaling, blockage and inter-plate leakage during use. These failures will not only reduce the heat exchange efficiency of the heat exchanger and cause energy loss, but may also lead to equipment failure and production line shutdown.

[0003] Due to the closed structure of the heat exchanger, its internal operating state parameters are difficult to obtain directly. Usually, its operating state can only be indirectly judged by measuring the state parameters of the fluid at the inlet and outlet of the heat exchanger. At present, the fault diagnosis of heat exchangers mainly relies on manual inspections and judgments based on experience. With the development of information technology and modern control technology, data-driven fault diagnosis methods have gradually become a research hotspot. This method collects data through sensors and uses artificial intelligence algorithms for fault identification, aiming to improve the accuracy of diagnosis and reduce economic costs.

[0004] Although data-driven methods have shown potential in improving the accuracy of fault diagnosis, existing data-driven methods still have some limitations. Since the data collected by sensors are usually large in dimension, there is redundancy in the data, which not only increases the complexity of data processing but also affects the efficiency of diagnosis. In addition, the features in high-dimensional data are often not independent of each other, which may cause information overlap between features and reduce the performance of the diagnosis model. Summary of the invention

[0005] The present invention proposes a heat exchanger fault diagnosis method and system based on data dimensionality reduction, which solves the data redundancy problem caused by large data dimension through feature extraction and dimensionality reduction, and improves the independence of features by learning the intrinsic representation of data, thereby improving the accuracy and efficiency of fault diagnosis.

[0006] To achieve the above object, the present invention provides a heat exchanger fault diagnosis method based on data dimension reduction, comprising the following steps:

[0007] Step S1: collecting operating data of the heat exchanger under various working conditions, and constructing a data set according to the operating data and the working conditions corresponding to the operating data;

[0008] Step S2: extracting features of data in the data set to obtain feature data, performing dimension reduction on the feature data to obtain a training set, and using the training set to train a classification model;

[0009] Step S3: Collect the operating data of the heat exchanger to be diagnosed, extract the features of the operating data of the heat exchanger to be diagnosed, reduce the dimensions of the features to obtain reduced-dimensional data, input the reduced-dimensional data into a trained classification model, and the classification model outputs the state and working condition of the heat exchanger to be diagnosed, so as to realize fault diagnosis of the heat exchanger.

[0010] Preferably, in step S1, the inlet flow rate, inlet and outlet temperature, inlet pressure and inlet and outlet pressure difference of the cold fluid and the hot fluid of the heat exchanger are collected by sensors, and the heat release of the hot fluid, the heat absorption of the cold fluid, the average temperature difference between the cold and hot fluids and the total heat transfer coefficient in the heat exchanger are calculated based on the collected data, and the inlet flow rate, inlet and outlet temperature, inlet pressure and inlet and outlet pressure difference of the cold fluid and the hot fluid, the heat release of the hot fluid, the heat absorption of the cold fluid, the average temperature difference between the cold and hot fluids, the total heat transfer coefficient, the temperature difference between the cold fluid and the hot fluid, and the difference between the heat release of the hot fluid and the heat absorption of the cold fluid are used as operating data.

[0011] Preferably, the expression for calculating the heat release of the hot fluid and the heat absorption of the cold fluid is:

[0012]

[0013] In the formula, Φ 1 is the heat release of the thermal fluid; Φ 2 is the heat absorbed by the cold fluid; q m1 and q m2 are the mass flow rates of hot and cold fluids respectively; C 1 , C 2 are the average specific heat capacities of hot and cold fluids at constant pressure within the inlet and outlet temperature ranges, respectively; t 1 'and t' 2 are the inlet temperatures of the hot fluid and the cold fluid respectively; t 1 ” and t 2 " represent the outlet temperatures of hot and cold fluids respectively.

[0014] Preferably, the expression for calculating the average temperature difference between the cold and hot fluids is:

[0015]

[0016] In the formula, Δt m Indicates the average temperature difference of the fluid heat transfer on both sides of the heat exchanger wall; t 1 'and t' 2 are the inlet temperatures of the hot fluid and the cold fluid respectively; t 1 ” and t 2 " represent the outlet temperatures of hot and cold fluids respectively.

[0017] Preferably, the expression for calculating the total heat transfer coefficient is:

[0018]

[0019] Where A is the heat transfer area based on the outer surface of the heat exchanger base tube; Φ m is the average heat flow of the heat exchanger; Δt m is the average temperature difference between the hot and cold fluids in the heat exchanger; Φ 1 is the heat released by the thermal fluid; Φ 2 is the amount of heat absorbed by the cold fluid.

[0020] Preferably, in step S2, a stacked denoising autoencoder SDAE is used to perform feature extraction and dimensionality reduction processing on the data set.

[0021] Preferably, the classification model in step S2 is a random forest model.

[0022] The present invention also provides a heat exchanger fault diagnosis system based on data dimension reduction, which is implemented based on the above-mentioned heat exchanger fault diagnosis method based on data dimension reduction, and includes: a data acquisition module, a feature extraction and dimension reduction module, a fault classification module and a monitoring module;

[0023] The data acquisition module is used to collect operating data of the heat exchanger;

[0024] The feature extraction and dimensionality reduction module performs feature extraction and dimensionality reduction processing on the operation data collected by the data collection module, and converts the operation data into a low-dimensional feature representation;

[0025] The fault classification module identifies the fault type of the heat exchanger according to the feature data output by the feature extraction and dimension reduction module, thereby realizing the fault diagnosis of the heat exchanger;

[0026] The monitoring module is used to monitor the operating status of the heat exchanger in real time to ensure the stable operation of the heat exchanger system.

[0027] Preferably, the feature extraction and dimensionality reduction module uses a stacked denoising autoencoder SDAE to perform feature extraction and dimensionality reduction processing on the operating data collected by the data collection module, and the non-output layer of the stacked denoising autoencoder SDAE includes multiple hidden layers.

[0028] Preferably, the fault classification module uses a random forest model to implement fault diagnosis of the heat exchanger.

[0029] The benefits of the present invention include at least:

[0030] 1. By performing feature extraction and dimensionality reduction on the data set, key information can be effectively screened out from the original data, noise and redundancy in the data can be removed, and the model training efficiency and diagnostic accuracy can be improved. Using the reduced-dimensional feature data to train the classification model can reduce the time and computing resources required for model training, while avoiding the problem of overfitting;

[0031] 2. By training the model on data sets under various working conditions, the model's adaptability and generalization ability to different working conditions are improved, making the model more reliable in practical applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 A schematic diagram of a method flow of an embodiment of the present invention;

[0033] Figure 2 Schematic diagram of SDAE output results in an embodiment of the present invention;

[0034] Figure 3 Schematic diagram of a confusion matrix obtained after classifying a test set in an embodiment of the present invention. DETAILED DESCRIPTION

[0035] The following is a clear and complete description of the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work belong to the protection scope of the present invention.

[0036] like Figure 1 As shown, an embodiment of the present invention provides a heat exchanger fault diagnosis method based on data dimension reduction, comprising the following steps:

[0037] Step S1: collecting operating data of the heat exchanger under various operating conditions, and constructing a data set according to the collected operating data and the operating conditions corresponding to the operating data.

[0038] Specifically, several flow sensors, temperature sensors and pressure sensors are installed on the heat exchanger. The flow sensor is used to measure the flow of the hot fluid and the cold fluid of the heat exchanger, the temperature sensor is used to measure the temperature of the fluid before entering the heat exchanger and after leaving the heat exchanger, and the pressure sensor is used to measure the inlet pressure of the two fluids and the pressure difference between the inlet and outlet.

[0039] After setting up the sensor, experimental data were collected under various working conditions. The experimental data were measured by fixing the cooling water flow rate and adjusting the lubricating oil flow rate. According to GB / T 19700-2023, the flow rate of the heat exchanger should vary within the range of 0.1m / s-1.0m / s. Therefore, in the embodiment of the present invention, the lubricating oil flow rate of the heat exchanger was set to 0.3m / s, 0.5m / s and 0.7m / s, respectively, and the corresponding flow rates were 17.34m / s, 17.34m / s and 17.34m / s, respectively. 3 / h、28.9m 3 / h and 40.47m 3 / h. The three working conditions of the experiment are: oil flow rate of 17.34m 3 / h and cooling water flow rate is 35m 3 / h, the oil flow rate is 28.9m 3 / h and cooling water flow rate is 35m 3 / h, the oil flow rate is 40.47m 3 / h and cooling water flow rate is 35m 3 / h. Each working condition has four states: normal data, 5% inner tube bundle blockage, 10% inner tube bundle blockage, and 20% inner tube bundle blockage. 600 samples were collected in each state, and the specific composition of the constructed data set is shown in Table 1.

[0040] Table 1 Specific composition of the dataset

[0041]

[0042]

[0043] The heat release of the hot fluid, the heat absorption of the cold fluid, the average temperature difference between the hot and cold fluids and the total heat transfer coefficient in the heat exchanger are calculated based on the collected data. The inlet flow rate, inlet and outlet temperature, inlet pressure and inlet and outlet pressure difference of the cold and hot fluids, the heat release of the hot fluid, the heat absorption of the cold fluid, the average temperature difference between the hot and cold fluids, the total heat transfer coefficient, the temperature difference between the cold and hot fluids, and the difference between the heat release of the hot fluid and the heat absorption of the cold fluid are used as operating data to construct a data set.

[0044] Specifically, according to the inlet temperature and outlet temperature of the hot and cold fluids of the heat exchanger collected by the sensor, the heat release of the hot fluid and the heat absorption of the cold fluid of the heat exchanger are calculated, and the expression is:

[0045]

[0046] In the formula, Φ 1 is the heat released by the thermal fluid, in W; Φ 2 is the heat absorbed by the cold fluid, in W; q m1 and q m2are the mass flow rates of hot and cold fluids, in kg / s; C 1 , C 2 are the average constant-pressure specific heat capacities of the hot and cold fluids within the inlet and outlet temperature ranges, respectively, in J / (Kg·℃); t 1 'and t' 2 are the inlet temperatures of the hot fluid and the cold fluid, respectively, in °C; t 1 ” and t 2 " respectively represent the outlet temperatures of the hot and cold fluids, in °C.

[0047] The expression for calculating the average temperature difference between the cold and hot fluids is:

[0048]

[0049] In the formula, Δt m It represents the average temperature difference of the fluid heat transfer on both sides of the heat exchanger partition, in degrees Celsius.

[0050] The total heat transfer coefficient of the heat exchanger can be calculated based on the average heat flow of the heat exchanger and the average temperature difference calculated in the previous step, that is, the average heat transfer coefficient on the entire heat transfer surface of the heat exchanger, in units of w / m2·℃. The total heat transfer coefficient can be used to evaluate the heat transfer performance of the multi-layer solid wall of the heat exchanger. The expression for calculating the total heat transfer coefficient is:

[0051]

[0052] Where A is the heat transfer area based on the outer surface of the heat exchanger base tube, in m 2 ; Φ m is the average heat flow of the heat exchanger, in W.

[0053] In order to eliminate the impact of different data dimensions, it is also necessary to normalize the data set. The operation data of the heat exchanger collected by each sensor and the calculated data are normalized to the minimum and maximum values, and the data are normalized to between 0 and 1. The specific calculation method is as follows:

[0054]

[0055] In the formula, x max is the maximum value of the column where the data is located, x min is the minimum value of the column where the data is located, x is the original data, and x new Normalized data.

[0056] The dataset is divided into two parts: a test set and a training set, with the training set accounting for 2 / 3 and the test set accounting for 1 / 3.

[0057] Step S2: extract features from the data set to obtain feature data, perform dimension reduction on the feature data to obtain a training set, and use the training set to train a classification model.

[0058] Specifically, a stacked denoising autoencoder SDAE with four hidden layers and no output layer is constructed. The normalized training set is used as the input of SDAE. After SDAE processing, data dimensionality reduction can be achieved. By removing irrelevant or redundant features, reducing the dimensionality of the data can improve the generalization ability of the model and reduce the risk of overfitting. At the same time, it can reduce the amount of calculation during model training and prediction, making the algorithm run faster. The reduced dimensionality dataset occupies less storage space, and the data can be projected into two-dimensional or three-dimensional space for easy observation and analysis.

[0059] Figure 2 Visualization of the results of extracting features for SDAE and reducing the feature dimension to 2D. Figure 2 It can be seen that after the data is extracted by SDAE, the data features of different categories are quite different, indicating that the unsupervised SDAE model has the advantages of strong adaptive learning and feature extraction capabilities.

[0060] Since there are 17 data collected and calculated each time, the number of input layer nodes of the SDAE model is set to 17. According to historical experience, the number of nodes of the four hidden layers in the embodiment of the present invention is set to 10, 8, 4 and 2 respectively, the number of iterations of the four hidden layers is 1000, the learning rate is 0.06, and the denoising rate is 0.08. Using this SDAE with a network structure of [17, 10, 8, 4, 2], the data set can be reduced to 2 dimensions.

[0061] The feature data extracted by SDAE is divided into training set and test set again, where the training set and test set account for 2 / 3 and 1 / 3 of the total number of samples respectively. The data of the training set is used as the input of the random forest to train the random forest model. The trained random forest model is used to classify the test set data, and the classification accuracy is used to evaluate the classification performance of the model. The test set samples are classified according to the above method, and the classification accuracy is 98.96%. Figure 3 This is the confusion matrix obtained after classifying the test set in the embodiment of the present invention. Figure 3 It can be seen that the method proposed in the present invention has a very high accuracy rate, and compared with the method of manually observing data to judge faults, the method of the present invention can effectively improve the accuracy and efficiency of fault diagnosis, and help to ensure the stable operation and safe production of the heat exchanger system.

[0062] Step S3: Collect the operating data of the heat exchanger to be diagnosed, and input the operating data of the heat exchanger to be diagnosed into the SDAE model. The SDAE model can output the features of the reduced-dimensional operating data. The reduced-dimensional features are input into the trained random forest model. The random forest model can output the state and working condition of the heat exchanger to be diagnosed, so as to realize the fault diagnosis of the heat exchanger.

[0063] The present invention provides a heat exchanger fault diagnosis method based on data dimension reduction. The method uses the SDAE model to extract features and reduce the dimension of the data collected during the use of the target heat exchanger, and can obtain low-dimensional representations of different fault categories. The method of the present invention can establish a model without understanding the structure and operation mechanism of the heat exchanger, reducing the difficulty of modeling and the reliance on expert experience. Compared with the traditional method of manually observing data to determine the type of fault and prevent the occurrence of faults, it can reduce the manpower demand in industrial production, shorten the maintenance time after the fault occurs, and improve the accuracy of fault diagnosis.

[0064] An embodiment of the present invention also provides a heat exchanger fault diagnosis system based on data dimensionality reduction, which is implemented based on the above-mentioned heat exchanger fault diagnosis method based on data dimensionality reduction, and includes a data acquisition module, a feature extraction and dimensionality reduction module, a fault classification module and a monitoring module.

[0065] Among them, the data acquisition module is used to collect the operating data of the heat exchanger. The feature extraction and dimension reduction module performs feature extraction and dimension reduction processing on the operating data collected by the data acquisition module, and converts the operating data into a low-dimensional feature representation. The fault classification module identifies the fault type of the heat exchanger based on the feature data output by the feature extraction and dimension reduction module to realize the fault diagnosis of the heat exchanger. The monitoring module is used to monitor the operating status of the heat exchanger in real time to ensure the stable operation of the heat exchanger system.

[0066] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. Only the preferred embodiments of the present invention are expressed. The description is more specific and detailed, but it cannot be understood as limiting the scope of the patent of the present invention. As long as there is no contradiction in the combination of these technical features, they should be considered as the scope recorded in this specification.

[0067] It should be noted that, for those skilled in the art, several modifications and improvements can be made without departing from the concept of the present invention, and these modifications and improvements all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

Claims

1. A heat exchanger fault diagnosis method based on data dimensionality reduction, characterized in that: The following steps are involved: Step S1: collecting operating data of the heat exchanger under various operating conditions, and constructing a data set based on the operating data and the operating conditions corresponding to the operating data; Step S2: extracting features of the data in the data set to obtain feature data, performing dimensionality reduction on the feature data to obtain a training set, and using the training set to train a classification model; Step S3: Collect the operating data of the heat exchanger to be diagnosed, extract the features of the operating data of the heat exchanger to be diagnosed, reduce the dimensions of the features to obtain reduced-dimensionality data, input the reduced-dimensionality data into a trained classification model, and the classification model outputs the status and operating conditions of the heat exchanger to be diagnosed, thereby realizing fault diagnosis of the heat exchanger.

2. The heat exchanger fault diagnosis method based on data dimension reduction according to claim 1, characterized in that: In step S1, the inlet flow rate, inlet and outlet temperature, inlet pressure and inlet and outlet pressure difference of the cold fluid and the hot fluid of the heat exchanger are collected by sensors. The heat release of the hot fluid, the heat absorption of the cold fluid, the average temperature difference between the cold and hot fluids and the total heat transfer coefficient in the heat exchanger are calculated based on the collected data. The inlet flow rate, inlet and outlet temperature, inlet pressure and inlet and outlet pressure difference of the cold fluid and the hot fluid, the heat release of the hot fluid, the heat absorption of the cold fluid, the average temperature difference between the cold and hot fluids, the total heat transfer coefficient, the temperature difference between the cold fluid and the hot fluid, and the difference between the heat release of the hot fluid and the heat absorption of the cold fluid are used as operating data.

3. The heat exchanger fault diagnosis method based on data dimension reduction according to claim 2, characterized in that: The expression for calculating the heat release of the hot fluid and the heat absorption of the cold fluid is: Where Φ1 is the heat released by the hot fluid; Φ2 is the heat absorbed by the cold fluid; q m1 and q m2 are the mass flow rates of hot and cold fluids respectively; C1 and C2 are the average constant pressure specific heat capacities of hot and cold fluids within the inlet and outlet temperature ranges respectively; t1' and t'2 are the inlet temperatures of the hot fluid and the cold fluid respectively; t1" and t2" represent the outlet temperatures of the hot fluid and the cold fluid respectively.

4. The heat exchanger fault diagnosis method based on data dimension reduction according to claim 2, characterized in that: The expression for calculating the average temperature difference between the cold and hot fluids is: Where Δt m It represents the average temperature difference of the fluid heat transfer on both sides of the heat exchanger wall; t1' and t'2 are the inlet temperatures of the hot fluid and the cold fluid respectively; t1" and t2" represent the outlet temperatures of the hot and cold fluid respectively.

5. The heat exchanger fault diagnosis method based on data dimension reduction according to claim 2, characterized in that: The expression for calculating the overall heat transfer coefficient is: Where A is the heat transfer area based on the outer surface of the heat exchanger base tube; Φ m is the average heat flow of the heat exchanger; Δt m is the average temperature difference between the hot and cold fluids in the heat exchanger; Φ1 is the heat released by the hot fluid; Φ2 is the heat absorbed by the cold fluid.

6. The heat exchanger fault diagnosis method based on data dimension reduction according to claim 1, characterized in that: In step S2, the stacked denoising autoencoder (SDAE) is used to perform feature extraction and dimensionality reduction on the dataset.

7. The heat exchanger fault diagnosis method based on data dimension reduction according to claim 1, characterized in that: The classification model in step S2 is a random forest model.

8. A heat exchanger fault diagnosis system based on data dimensionality reduction, implemented based on a heat exchanger fault diagnosis method based on data dimensionality reduction according to any one of claims 1 to 7, characterized in that: include: Data acquisition module, feature extraction and dimensionality reduction module, fault classification module and monitoring module; The data acquisition module is used to collect operating data of the heat exchanger; The feature extraction and dimensionality reduction module performs feature extraction and dimensionality reduction processing on the operation data collected by the data collection module, and converts the operation data into a low-dimensional feature representation; The fault classification module identifies the fault type of the heat exchanger based on the feature data output by the feature extraction and dimensionality reduction module, thereby realizing fault diagnosis of the heat exchanger; The monitoring module is used to monitor the operating status of the heat exchanger in real time to ensure the stable operation of the heat exchanger system.

9. The heat exchanger fault diagnosis system based on data dimension reduction according to claim 8, characterized in that: The feature extraction and dimensionality reduction module uses a stacked denoising autoencoder SDAE to perform feature extraction and dimensionality reduction processing on the operating data collected by the data collection module, and the non-output layer of the stacked denoising autoencoder SDAE includes multiple hidden layers.

10. The heat exchanger fault diagnosis system based on data dimension reduction according to claim 8, characterized in that: The fault classification module uses a random forest model to implement fault diagnosis of the heat exchanger.

Citation Information

Patent Citations

  • Heat exchanger early fault diagnosis method based on BP neural network

    CN110779745A

  • Visual intelligent monitoring system and method for ash deposition corrosion wear state of medium-low temperature smoke heat exchange equipment

    CN111855547A

  • Heat exchanger industrial calibration device

    CN112504716A

  • Power system disturbance classification method

    CN112633315A

  • Dynamic dirt monitoring method and system for low-temperature coal economizer of coal-fired boiler

    CN112923349A

Cited By

  • Heat exchanger fault diagnosis and prediction maintenance system and method based on data analysis

    CN119377793A