Nuclear power plant heat exchanger life prediction method and device and related equipment

Abstract

The invention relates to the field of nuclear power safety detection, in particular to a nuclear power plant heat exchanger service life prediction method and device and related equipment. The method comprises the following steps: acquiring a pre-constructed life prediction model; the service life prediction model is pre-constructed through the following steps that historical equipment degradation data of similar heat exchangers corresponding to a target heat exchanger are obtained; performing health feature extraction on the historical equipment degradation data to obtain a plurality of health feature dimensions; based on the plurality of health feature dimensions, constructing a composite health index and a failure threshold of the composite health index; based on the composite health index and the failure threshold value, a service life prediction model corresponding to the target heat exchanger is constructed; acquiring multiple pieces of target equipment data acquired by the target measuring point; and the multiple pieces of target equipment data are input into a life prediction model, life prediction is conducted through the life prediction model, and the target remaining life of the target heat exchanger is obtained. The embodiment of the invention can improve the accuracy of predicting the life of the heat exchanger of the nuclear power plant.

Description

Technical Field

[0001] This application relates to the field of nuclear power safety detection, and particularly to a method, device and related equipment for predicting the service life of heat exchangers in nuclear power plants. Background Art

[0002] The heat exchanger is one of the crucial equipment in a nuclear power plant. Its main function is to transfer the heat in the reactor coolant to the cooling medium in the secondary loop through heat exchange to ensure the safe operation of the reactor system. As the core component of the nuclear power plant's thermal system, the performance of the heat exchanger directly affects the operation efficiency and safety of the nuclear power plant. Therefore, the reliable operation of the heat exchanger is an important guarantee for the safe and stable operation of the nuclear power plant.

[0003] In the related art, the life prediction of heat exchangers in nuclear power plants mainly relies on a regular preventive maintenance strategy based on empirical formulas and statistical data. Although this method is simple and easy to implement, it has certain limitations: on the one hand, it may lead to over-maintenance, increasing the equipment maintenance cost; on the other hand, there may also be insufficient maintenance, affecting the safe operation of the equipment and even causing serious safety accidents. Summary of the Invention

[0004] This application aims to solve at least one of the technical problems existing in the prior art. For this reason, a method, device and related equipment for predicting the service life of heat exchangers in nuclear power plants provided by this application can improve the accuracy of predicting the service life of heat exchangers in nuclear power plants.

[0005] In a first aspect, an embodiment of this application provides a method for predicting the service life of a heat exchanger in a nuclear power plant, and the method includes:

[0006] Obtain a pre-constructed life prediction model; wherein, the life prediction model is pre-constructed by the following steps:

[0007] Obtain the historical equipment degradation data of the same type of heat exchangers corresponding to the target heat exchanger;

[0008] Extract health features from the historical equipment degradation data to obtain multiple health feature dimensions;

[0009] Based on multiple health feature dimensions, construct a composite health index and a failure threshold of the composite health index;

[0010] Based on the composite health index and the failure threshold, construct the life prediction model corresponding to the target heat exchanger;

[0011] Obtain multiple target equipment data collected by a target measuring point; wherein, the target measuring point is configured on the target heat exchanger for collecting the equipment data of the target heat exchanger;

[0012] Input the multiple pieces of target device data into the life prediction model to perform life prediction through the life prediction model, and obtain the target remaining life of the target heat exchanger.

[0013] In some embodiments, each of the health feature dimensions includes multiple sample data. Based on the multiple health feature dimensions, constructing a composite health index and a failure threshold of the composite health index includes:

[0014] Construct a health index decision matrix according to the multiple health feature dimensions and the multiple sample data;

[0015] Based on the health index decision matrix, calculate the dimension weight corresponding to each health feature dimension;

[0016] Calculate the dimension correlation coefficient corresponding to each health feature dimension according to a preset multi-dimensional health feature reference sequence;

[0017] Construct a composite health index of the target heat exchanger according to the dimension weight and the correlation coefficient corresponding to each health feature dimension;

[0018] Based on the historical equipment degradation data of the same type of heat exchanger at the time of failure, determine the failure threshold of the composite health index.

[0019] In some embodiments, the calculating the dimension weight corresponding to each health feature dimension based on the health index decision matrix includes:

[0020] Calculate the feature probability corresponding to each health feature dimension according to the health index decision matrix;

[0021] Calculate the dimension entropy value corresponding to the health feature dimension according to the feature probability corresponding to each health feature dimension;

[0022] Calculate the dimension weight corresponding to the health feature dimension according to the dimension entropy value corresponding to each health feature dimension.

[0023] In some embodiments, the calculating the dimension correlation coefficient corresponding to each health feature dimension according to a preset multi-dimensional health feature reference sequence includes:

[0024] Calculate the difference value between each sample data in each health feature dimension and the corresponding data in the reference sequence;

[0025] Determine the maximum difference value and the minimum difference value among all the difference values;

[0026] Determine the dimension correlation coefficient corresponding to each health feature dimension according to the difference value, the maximum difference value, the minimum difference value, and a preset resolution coefficient.

[0027] In some embodiments, constructing the composite health index of the target heat exchanger according to the dimension weight and the correlation coefficient corresponding to each health feature dimension includes:

[0028] Calculate the weighted correlation coefficient corresponding to the health feature dimension according to the dimension weight and the dimension correlation coefficient corresponding to each health feature dimension;

[0029] Perform a weighted combination of the weighted correlation coefficients corresponding to all the health feature dimensions to obtain the composite health index of the target heat exchanger.

[0030] In some embodiments, constructing the life prediction model corresponding to the target heat exchanger based on the composite health index and the failure threshold includes:

[0031] Construct a Wiener degradation process model according to the composite health index;

[0032] Based on the Wiener degradation process model and the failure threshold, construct a remaining life expression; wherein, the remaining life expression is used to represent the time required for the target heat exchanger to reach the failure threshold from the current state;

[0033] Construct a probability density function of the remaining life according to the remaining life expression to obtain the life prediction model corresponding to the target heat exchanger.

[0034] In some embodiments, the process of pre-determining the target measurement points includes:

[0035] Based on the historical equipment degradation data, establish a multi-source observation model between the equipment state of the target heat exchanger and the sensor data corresponding to multiple sensor measurement points; wherein, the multi-source observation model is used to describe the mapping relationship between the equipment state and multiple sensor data;

[0036] Based on the multi-source observation model, construct a state evaluation coefficient of the target heat exchanger; wherein, the state evaluation coefficient is used to characterize the accuracy of the sensor data corresponding to each sensor measurement point in evaluating the equipment state;

[0037] Based on each state evaluation coefficient and a preset resource consumption coefficient, establish a measurement point evaluation function model; wherein, the measurement point evaluation function model is used to evaluate the comprehensive performance of each sensor measurement point;

[0038] Based on the measurement point evaluation function model, select multiple target measurement points from multiple candidate sensor measurement points.

[0039] In some embodiments, constructing the state evaluation coefficient of the target heat exchanger based on the multi-source observation model includes:

[0040] Determine the observation equation for each candidate sensor measurement point according to the multi-source observation model; wherein, the observation equation is used to describe the relationship between single-sensor data and the device state;

[0041] Based on the observation equation, construct the state mapping function for each candidate sensor measurement point; wherein, the state mapping function is used to map sensor data to the device state;

[0042] Calculate the state evaluation error corresponding to each candidate sensor measurement point according to the historical device degradation data;

[0043] Based on the state evaluation error, determine the state evaluation coefficient corresponding to each candidate sensor measurement point.

[0044] In some embodiments, establishing the measurement point evaluation function model based on each state evaluation coefficient and a preset resource consumption coefficient includes:

[0045] Construct the selection state space and selection action space for the multiple sensor measurement points; wherein, the selection state space includes the state information of whether each sensor measurement point is selected, and the selection action space includes the selection operation of each sensor measurement point;

[0046] Based on the selection state space and the selection action space, construct a value evaluation function; wherein, the value evaluation function is used to characterize the expected cumulative reward for performing the selection action in the current selection state;

[0047] Construct a reward function according to the state evaluation coefficient corresponding to each sensor measurement point and the preset resource consumption coefficient;

[0048] Based on the value evaluation function and the reward function, optimize the selection strategy of the sensor measurement points to obtain the measurement point evaluation function model.

[0049] In some embodiments, performing life prediction through the life prediction model to obtain the target remaining life of the target heat exchanger includes:

[0050] Perform state estimation on the target device data based on the Kalman filter algorithm to obtain a state estimation result;

[0051] Perform life prediction on the state estimation result through the life prediction model to obtain the target remaining life of the target heat exchanger.

[0052] Second aspect, an embodiment of the present application provides a device for predicting the life of a heat exchanger in a nuclear power plant, including:

[0053] A first acquisition module, configured to acquire a pre-constructed life prediction model; wherein, the life prediction model is pre-constructed by the following steps:

[0054] Acquire historical equipment degradation data of the same type of heat exchanger corresponding to the target heat exchanger;

[0055] Extract health features from the historical equipment degradation data to obtain multiple health feature dimensions;

[0056] Based on multiple health feature dimensions, construct a composite health index and a failure threshold of the composite health index;

[0057] Based on the composite health index and the failure threshold, construct the life prediction model corresponding to the target heat exchanger;

[0058] A second acquisition module, configured to acquire a plurality of target device data collected by a target measuring point; wherein, the target measuring point is configured on the target heat exchanger to collect device data of the target heat exchanger;

[0059] An input module, configured to input the plurality of target device data into the life prediction model to perform life prediction through the life prediction model, and obtain the target remaining life of the target heat exchanger.

[0060] Third aspect, an embodiment of the present application provides an electronic device, including: a memory, a processor, the memory stores a computer program, and when the processor executes the computer program, it implements the method for predicting the life of a heat exchanger in a nuclear power plant according to any one of the embodiments in the first aspect of the present application.

[0061] Fourth aspect, an embodiment of the present application provides a computer-readable storage medium, the storage medium stores a program, and when the program is executed by a processor, it implements the method for predicting the life of a heat exchanger in a nuclear power plant according to any one of the embodiments in the first aspect of the present application.

[0062] The method for predicting the life of a heat exchanger in a nuclear power plant according to the embodiments of the present application has at least the following beneficial effects:

[0063] The method for predicting the service life of a heat exchanger in a nuclear power plant according to an embodiment of the present application includes: obtaining a pre-constructed service life prediction model; wherein, the service life prediction model is pre-constructed by the following steps: obtaining historical equipment degradation data of the same type of heat exchanger corresponding to the target heat exchanger; extracting health features from the historical equipment degradation data to obtain multiple health feature dimensions; constructing a composite health index and a failure threshold of the composite health index based on the multiple health feature dimensions; constructing the service life prediction model corresponding to the target heat exchanger based on the composite health index and the failure threshold; obtaining multiple target equipment data collected by a target measurement point; wherein, the target measurement point is configured on the target heat exchanger for collecting equipment data of the target heat exchanger; inputting the multiple target equipment data into the service life prediction model to perform service life prediction through the service life prediction model, so as to obtain the target remaining service life of the target heat exchanger.

[0064] Aiming at the technical problem that the service life prediction of heat exchangers in nuclear power plants in the background technology overly relies on empirical formulas and statistical data, the present application provides a method for predicting the service life of a heat exchanger based on a composite health index. This method first obtains historical equipment degradation data of the same type of heat exchanger and extracts health features from these data to obtain multiple health feature dimensions, thereby making full use of existing operating experience and overcoming the limitations of relying solely on empirical formulas. On this basis, the present application constructs a composite health index and its failure threshold based on the extracted multiple health feature dimensions. This composite health index can not only more comprehensively reflect the health status of the equipment, but also reduce the misjudgment risk caused by a single feature by considering the combined effect of multi-dimensional features. Subsequently, the present application combines the composite health index and the failure threshold to construct a dedicated service life prediction model, realizing the transformation from static empirical judgment to dynamic data-driven. Finally, by configuring measurement points on the target heat exchanger to collect real-time data and inputting these data into the pre-constructed prediction model, the dynamic prediction of the remaining service life of the equipment is realized, and the change trend of the equipment status is captured in a timely manner. The organic combination of the above technical means overcomes the problems of excessive maintenance or insufficient maintenance that may occur in the prior art, significantly improves the accuracy of the service life prediction of heat exchangers in nuclear power plants, provides more reliable data support for equipment maintenance decisions, and thus ensures the safe and stable operation of nuclear power plants.

[0065] Additional aspects and advantages of the present application will be given in part in the following description, become apparent in part from the following description, or be understood through the practice of the present application. Description of the Drawings

[0066] The above and / or additional aspects and advantages of the present application will become apparent and be readily understood from the description of the embodiments in conjunction with the following drawings, wherein:

[0067] Figure 1Flow chart of an optional method for predicting the service life of a heat exchanger in a nuclear power plant provided by an embodiment of the present application;

[0068] Figure 2 Flow chart of another optional method for predicting the service life of a heat exchanger in a nuclear power plant provided by an embodiment of the present application;

[0069] Figure 3 Flow chart of another optional method for predicting the service life of a heat exchanger in a nuclear power plant provided by an embodiment of the present application;

[0070] Figure 4 Flow chart of another optional method for predicting the service life of a heat exchanger in a nuclear power plant provided by an embodiment of the present application;

[0071] Figure 5 Flow chart of another optional method for predicting the service life of a heat exchanger in a nuclear power plant provided by an embodiment of the present application;

[0072] Figure 6 Flow chart of another optional method for predicting the service life of a heat exchanger in a nuclear power plant provided by an embodiment of the present application;

[0073] Figure 7 Flow chart of another optional method for predicting the service life of a heat exchanger in a nuclear power plant provided by an embodiment of the present application;

[0074] Figure 8 Flow chart of another optional method for predicting the service life of a heat exchanger in a nuclear power plant provided by an embodiment of the present application;

[0075] Figure 9 Flow chart of another optional method for predicting the service life of a heat exchanger in a nuclear power plant provided by an embodiment of the present application;

[0076] Figure 10 Flow chart of another optional method for predicting the service life of a heat exchanger in a nuclear power plant provided by an embodiment of the present application;

[0077] Figure 11 Flow chart of another optional method for predicting the service life of a heat exchanger in a nuclear power plant provided by an embodiment of the present application;

[0078] Figure 12 Schematic diagram of a device for predicting the service life of a heat exchanger in a nuclear power plant provided by an embodiment of the present application;

[0079] Figure 13 Schematic diagram of the hardware structure of an electronic device provided by an embodiment of the present application. Detailed implementation manners

[0080] Embodiments of the present application will be described in detail below. Examples of the embodiments are shown in the accompanying drawings, where like or similar reference numerals denote like or similar elements or elements having like or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary only for explaining the present application and should not be construed as limiting the present application.

[0081] In the description of the present application, the meaning of "a number of" is one or more, the meaning of "a plurality of" is two or more, and understandings such as "greater than", "less than", "exceeding", etc. do not include the present number, and understandings such as "above", "below", "within", etc. include the present number. If there is a description of "first" and "second", it is only for the purpose of distinguishing technical features and should not be construed as indicating or implying relative importance or implicitly indicating the quantity of the indicated technical features or implicitly indicating the sequence of the indicated technical features.

[0082] In the description of the present application, it should be understood that for the orientation description, such as the orientation or positional relationship indicated by "upper", "lower", "left", "right", "front", "rear", etc. is based on the orientation or positional relationship shown in the accompanying drawings. It is only for the convenience of describing the present application and simplifying the description, rather than indicating or implying that the indicated device or element must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as limiting the present application.

[0083] In the description of this specification, the description with reference to terms such as "one embodiment", "some embodiments", "schematic embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described can be combined in a suitable manner in any one or more embodiments or examples.

[0084] In the description of the present application, it should be noted that unless otherwise clearly defined, words such as "set", "installed", "connected", etc. should be understood in a broad sense, and those skilled in the art can reasonably determine the specific meanings of the above words in the present application in combination with the specific content of the technical solution. In addition, the identification of specific steps below does not represent a limitation on the step sequence and execution logic. The execution sequence and execution logic between each step should be understood and inferred with reference to the content described in the embodiments.

[0085] The heat exchanger is one of the crucial equipment in a nuclear power plant. Its main function is to transfer the heat in the reactor coolant to the cooling medium in the secondary circuit through heat exchange, ensuring the safe operation of the reactor system. In the thermodynamic system of a nuclear power plant, the heat exchanger plays a core role, responsible for efficiently and safely transferring the heat generated by the reactor to the secondary circuit, and then driving the steam turbine to generate electricity. Due to the special operating environment of the nuclear power plant, the heat exchanger not only needs to have high-efficient heat exchange capacity, but also must have extremely high reliability and safety to cope with extreme working conditions such as high temperature, high pressure, and strong radiation.

[0086] In a nuclear power plant, the heat exchanger is usually divided into a primary side and a secondary side. The primary side is in direct contact with the reactor coolant and is responsible for transferring the heat generated by the reactor to the cooling medium on the secondary side. The secondary side then transfers the heat to the steam generator through the cooling medium (usually water), and then drives the steam turbine to generate electricity. Since the primary side is in direct contact with the reactor coolant, the heat exchanger bears extremely high temperature and pressure during operation, and at the same time has to cope with corrosive substances and radioactive substances that may exist in the coolant. Therefore, the material selection, structural design, and operation and maintenance of the heat exchanger are all crucial.

[0087] The performance of the heat exchanger directly affects the operation efficiency and safety of the nuclear power plant. If the heat exchanger fails or its performance deteriorates, it may lead to the failure to transfer the heat of the reactor coolant in time, resulting in serious consequences such as reactor overheating and pressure increase, and even may trigger a nuclear accident. Therefore, the reliable operation of the heat exchanger is an important guarantee for the safe and stable operation of the nuclear power plant. To ensure the long-term stable operation of the heat exchanger, the nuclear power plant usually conducts regular maintenance and life prediction on the heat exchanger to timely detect potential problems and take corresponding maintenance measures.

[0088] However, existing heat exchanger life prediction methods mainly rely on regular preventive maintenance strategies based on empirical formulas and statistical data. Although this method is simple and easy to implement, it has certain limitations. First, regular preventive maintenance strategies are usually based on fixed time intervals or empirical formulas and cannot be adjusted according to the actual operating status of the equipment. This may lead to premature maintenance of the equipment before it reaches the state that requires maintenance, resulting in over-maintenance. Over-maintenance not only increases the cost of equipment maintenance but also may lead to premature replacement of the equipment, wasting resources. Second, since the regular maintenance strategy cannot monitor the actual state of the equipment in real time, it may lead to serious degradation or failure of the equipment before reaching the maintenance cycle, but the maintenance is not carried out in time, resulting in under-maintenance. Under-maintenance may cause the equipment to malfunction during operation, affecting the normal operation of the nuclear power plant and even triggering serious safety accidents. In addition, existing life prediction methods mainly rely on empirical formulas and statistical data and cannot accurately reflect the actual degradation process of the equipment. Especially under complex working conditions, the degradation process of the equipment may be affected by various factors, resulting in inaccurate prediction results. The lack of accurate life prediction may lead to unreasonable maintenance plans for the equipment, unable to effectively prevent equipment failures, and affecting the operation safety of the nuclear power plant.

[0089] Based on this, the present application provides a heat exchanger life prediction method based on composite health indicators. This method first obtains the historical equipment degradation data of similar heat exchangers and extracts health characteristics from these data to obtain multiple health characteristic dimensions, thereby making full use of the existing operating experience and overcoming the limitations of relying only on empirical formulas. On this basis, the present application constructs a composite health indicator and its failure threshold based on the extracted multiple health characteristic dimensions. This composite health indicator can not only more comprehensively reflect the health state of the equipment but also reduce the misjudgment risk brought by a single characteristic by considering the combined effect of multi-dimensional characteristics. Subsequently, the present application combines the composite health indicator and the failure threshold to construct a special life prediction model, realizing the transformation from static empirical judgment to dynamic data-driven. Finally, by configuring measurement points on the target heat exchanger to collect real-time data and inputting these data into the pre-constructed prediction model, the dynamic prediction of the remaining life of the equipment is realized, and the change trend of the equipment state is captured in time. The organic combination of the above technical means overcomes the problems of over-maintenance or under-maintenance that may occur in the prior art, significantly improves the accuracy of heat exchanger life prediction in nuclear power plants, provides more reliable data support for equipment maintenance decisions, and thus ensures the safe and stable operation of nuclear power plants.

[0090] Please refer to Figure 1 , a nuclear power plant heat exchanger life prediction method provided by an embodiment of the present invention, the method may include, but is not limited to, the following steps 101 to step 103:

[0091] Step 101, obtain a pre-built life prediction model.

[0092] Please refer to Figure 2 , in some embodiments, the pre-building process of the life prediction model in Step 101 may include, but is not limited to, the following Steps 201 to 204:

[0093] Step 201, obtain the historical equipment degradation data of the same type of heat exchanger corresponding to the target heat exchanger.

[0094] Step 202, perform health feature extraction on the historical equipment degradation data to obtain multiple health feature dimensions.

[0095] Step 203, based on multiple health feature dimensions, construct a composite health index and a failure threshold of the composite health index.

[0096] Step 204, based on the composite health index and the failure threshold, construct a life prediction model corresponding to the target heat exchanger.

[0097] Step 102, obtain multiple target equipment data collected by the target measuring point; wherein, the target measuring point is configured on the target heat exchanger for collecting the equipment data of the target heat exchanger.

[0098] Step 103, input the multiple target equipment data into the life prediction model to perform life prediction through the life prediction model, and obtain the target remaining life of the target heat exchanger.

[0099] In Step 101 of some embodiments, the life prediction model is the core part of this method, and its pre-building process includes multiple steps.

[0100] In Step 201 of some embodiments, obtain the historical equipment degradation data of the same type of heat exchanger corresponding to the target heat exchanger. These historical data include the performance parameters, degradation characteristics, and final failure data of the same type of heat exchanger at different operation stages. The historical equipment degradation data is the basis for constructing the life prediction model. Through these data, the degradation law of the heat exchanger can be analyzed, providing a basis for subsequent health feature extraction and model construction. The historical data of the same type of heat exchanger can reflect the degradation trend of the equipment under similar working conditions, providing a reference for predicting the life of the target heat exchanger.

[0101] In Step 202 of some embodiments, the health feature dimension refers to the parameters closely related to the performance of the heat exchanger, such as temperature, pressure, flow rate, vibration, etc. By extracting these features, the most valuable information for life prediction can be screened out from a large amount of data, avoiding the interference of redundant data, and improving the prediction accuracy of the model. The process of health feature extraction usually includes steps such as data preprocessing, feature selection, and feature weighting to ensure that the extracted features can comprehensively reflect the health state of the equipment.

[0102] In step 203 of some embodiments, the composite health index is obtained by weighted combination of multiple single health features, and can comprehensively reflect the health state of the heat exchanger. The construction of the composite health index usually adopts the grey relational entropy method, and by calculating the weights and correlation coefficients of each health feature dimension, multiple features are synthesized into one index. The failure threshold is determined according to the historical data of similar heat exchangers at the time of failure, and is used to judge whether the heat exchanger reaches the failure state. The determination of the failure threshold is the key to life prediction, which directly affects the accuracy of the prediction result.

[0103] Please refer to Figure 3 , in some embodiments, step 203 may include, but is not limited to, the following steps 301 to 305:

[0104] Step 301, construct a health index decision matrix according to multiple health feature dimensions and multiple sample data.

[0105] Step 302, calculate the dimension weight corresponding to each health feature dimension based on the health index decision matrix.

[0106] Step 303, calculate the dimension correlation coefficient corresponding to each health feature dimension according to the preset multi-dimensional health feature reference sequence.

[0107] Step 304, construct the composite health index of the target heat exchanger according to the dimension weight and correlation coefficient corresponding to each health feature dimension.

[0108] Step 305, determine the failure threshold of the composite health index based on the historical equipment degradation data of similar heat exchangers at the time of failure.

[0109] In step 301 of some embodiments, each health feature dimension includes multiple sample data. First, construct a health index decision matrix according to multiple health feature dimensions and multiple sample data. The health feature dimension refers to the parameters closely related to the performance of the heat exchanger, such as temperature, pressure, flow rate, vibration, etc. Each health feature dimension contains multiple sample data, which reflect the operating state of the heat exchanger at different time points. The health index decision matrix is a two-dimensional matrix, where each row represents a sample and each column represents a health feature dimension. By constructing the health index decision matrix, the sample data of multiple health feature dimensions can be organized, providing a data basis for subsequent weight calculation and correlation coefficient calculation.

[0110] Exemplarily, assuming there are m sample data and n health feature dimensions, construct a health index decision matrix X=(x ij ), where x ij represents the data of the i-th sample data on the j-th health feature dimension.

[0111]

[0112] In some embodiments, it is also necessary to standardize the data because the dimensions of different health feature dimensions are very likely to be different. Therefore, the data should be standardized to eliminate the influence of dimensions. The min-max standardization method can be adopted:

[0113]

[0114] where min(x j ) and max(x j ) are the minimum and maximum values of the health feature dimension C j , respectively.

[0115] In step 302 of some embodiments, the dimension weight reflects the importance of each health feature dimension in the composite health index. The process of calculating the dimension weight generally includes the following steps: First, calculate the feature probability of each health feature dimension, that is, the proportion of the sample data of each dimension in the total sample data. Then, calculate the entropy value of each health feature dimension according to the feature probability. The entropy value is used to measure the amount of information contained in the dimension. Finally, calculate the weight of each health feature dimension according to the entropy value. The smaller the entropy value of a dimension, the greater its weight, indicating that the dimension is more important in the composite health index.

[0116] Please refer to Figure 4 , in some embodiments, step 302 may include, but is not limited to, the following steps 401 to 403:

[0117] Step 401: Calculate the feature probability corresponding to each health feature dimension according to the health index decision matrix.

[0118] Step 402: Calculate the dimension entropy value corresponding to the health feature dimension according to the feature probability corresponding to each health feature dimension.

[0119] Step 403: Calculate the dimension weight corresponding to the health feature dimension according to the dimension entropy value corresponding to each health feature dimension.

[0120] In step 401 of some embodiments, first calculate the feature probability corresponding to each health feature dimension according to the health index decision matrix. The health index decision matrix is a two-dimensional matrix, where each row represents a sample and each column represents a health feature dimension. The feature probability refers to the proportion of the sample data of each health feature dimension in the total sample data. By calculating the feature probability, the distribution of each health feature dimension in the overall data can be understood, providing a basis for the subsequent entropy value calculation. Among them, the calculation formula of the feature probability p ij is:

[0121]

[0122] Among them, x' ij is the standardized data, and m is the number of samples. The feature probability p ij represents the proportion of the j-th healthy feature dimension in the i-th sample.

[0123] In step 402 of some embodiments, according to the feature probability corresponding to each healthy feature dimension, the dimension entropy value corresponding to the healthy feature dimension is calculated. The entropy value is used to measure the amount of information contained in each healthy feature dimension. The larger the entropy value, the more uniform the data distribution of this dimension and the less the amount of information; the smaller the entropy value, the more uneven the data distribution of this dimension and the more the amount of information. By calculating the entropy value, the discrimination ability of each healthy feature dimension in life prediction can be evaluated. The entropy value H j has the following calculation formula:

[0124]

[0125] Among them, p ij is the feature probability of the j-th healthy feature dimension in the i-th sample, and m is the number of samples. The entropy value HH j is used to measure the effective information amount of the index. Suppose the values of a certain index vary relatively greatly among all samples, then its entropy value will be smaller, indicating that this index has a stronger discrimination ability.

[0126] In step 403 of some embodiments, according to the dimension entropy value corresponding to each healthy feature dimension, the dimension weight corresponding to the healthy feature dimension is calculated. The dimension weight reflects the importance of each healthy feature dimension in the composite health index. The smaller the entropy value of a dimension, the greater its weight, indicating that the importance of this dimension in the composite health index is higher. By calculating the dimension weight, the contribution degree of each healthy feature dimension to the heat exchanger life prediction can be determined. The dimension weight w j has the following calculation formula:

[0127]

[0128] Among them, H j is the entropy value of the j-th healthy feature dimension, and n is the number of healthy feature dimensions. The weight w j reveals the relative importance of each index in life prediction. The smaller the entropy value of an index, the greater its weight.

[0129] Through steps 401 to 403, by calculating the characteristic probability, entropy value, and weight, the importance of each health characteristic dimension in the composite health index can be determined. The characteristic probability reflects the distribution of each health characteristic dimension in the overall data, the entropy value measures the amount of information of each health characteristic dimension, and the weight reflects the contribution degree of each health characteristic dimension to life prediction. These steps provide a scientific basis for constructing the composite health index and significantly improve the accuracy and reliability of life prediction.

[0130] In step 303 of some embodiments, the reference sequence is the health characteristic data in the ideal state and is used to compare with the actual health characteristic data. The process of calculating the dimension correlation coefficient includes the following steps: First, calculate the difference value between the sample data in each health characteristic dimension and the corresponding data of the reference sequence. Then, determine the maximum difference value and the minimum difference value among all the difference values. Finally, calculate the correlation coefficient of each health characteristic dimension according to the difference value, the maximum difference value, the minimum difference value, and the preset discrimination coefficient. The correlation coefficient reflects the degree of closeness between the actual health characteristic data and the ideal state. The larger the correlation coefficient, the better the health state of this dimension.

[0131] Please refer to Figure 5 , in some embodiments, step 303 may include, but is not limited to, the following steps 501 to 503:

[0132] Step 501, calculate the difference value between the sample data in each health characteristic dimension and the corresponding data of the reference sequence.

[0133] Step 502, determine the maximum difference value and the minimum difference value among all the difference values.

[0134] Step 503, determine the dimension correlation coefficient corresponding to each health characteristic dimension according to the difference value, the maximum difference value, the minimum difference value, and the preset discrimination coefficient.

[0135] In step 501 of some embodiments, first calculate the difference value between the sample data in each health characteristic dimension and the corresponding data of the reference sequence. The reference sequence is the health characteristic data in the ideal state and is used to compare with the actual health characteristic data. The difference value reflects the deviation between the actual data and the ideal state. By calculating the difference value, the gap between the actual data and the ideal state of each health characteristic dimension can be quantified, providing a basis for the subsequent calculation of the correlation coefficient. The difference value Δ ij The calculation formula is:

[0136]

[0137] where x′ ij is the standardized data, is the ideal value of the j-th healthy feature dimension in the reference sequence. The difference value Δ ij represents the gap between the i-th sample and the ideal state in the j-th healthy feature dimension.

[0138] In step 502 of some embodiments, the maximum difference value and the minimum difference value among all the difference values are determined. The maximum difference value and the minimum difference value are important parameters for calculating the correlation coefficient, which reflect the maximum and minimum deviations between all sample data and the reference sequence. By determining the maximum difference value and the minimum difference value, a reference range can be provided for the subsequent calculation of the correlation coefficient. The maximum difference value and the minimum difference value are respectively expressed as:

[0139]

[0140] In step 503 of some embodiments, according to the difference value, the maximum difference value, the minimum difference value, and a preset resolution coefficient, the dimension correlation coefficient corresponding to each healthy feature dimension is determined. The correlation coefficient reflects the degree of proximity between the actual healthy feature data and the ideal state. The larger the correlation coefficient, the better the healthy state of the dimension. By calculating the correlation coefficient, the degree of proximity between each healthy feature dimension and the ideal state can be quantified, thereby providing a basis for constructing a composite health index. The correlation coefficient ξ ij is calculated by the following formula:

[0141]

[0142] where ρ is the resolution coefficient, usually taking a value of 0.5. The correlation coefficient ξ ij represents the degree of proximity between the i-th sample and the ideal state in the j-th healthy feature dimension. The larger the value of ξ ij the better the healthy state of the sample in this dimension.

[0143] Through steps 501 to 503, by calculating the difference value, determining the maximum difference value and the minimum difference value, and calculating the correlation coefficient, the degree of proximity between each healthy feature dimension and the ideal state can be quantified. The difference value reflects the deviation between the actual data and the ideal state, the maximum difference value and the minimum difference value provide a reference range, and the correlation coefficient quantifies the healthy state of each healthy feature dimension. These steps provide a scientific basis for constructing a composite health index, significantly improving the accuracy and reliability of life prediction.

[0144] In step 304 of some embodiments, the composite health index is obtained by weighted combination of multiple single health characteristics, and can comprehensively reflect the health state of the heat exchanger. Specifically, first, according to the weight and correlation coefficient of each health characteristic dimension, the weighted correlation coefficient of this dimension is calculated. Then, the weighted correlation coefficients of all health characteristic dimensions are weighted combined to obtain the composite health index of the target heat exchanger. The composite health index can comprehensively consider multiple degradation characteristics, avoid the limitations of a single index, and improve the accuracy of life prediction.

[0145] Please refer to Figure 6 , in some embodiments, step 304 may include, but is not limited to, the following steps 601 to step 602:

[0146] Step 601, according to the dimension weight and dimension correlation coefficient corresponding to each health characteristic dimension, calculate the weighted correlation coefficient corresponding to the health characteristic dimension.

[0147] Step 602, perform a weighted combination of the weighted correlation coefficients corresponding to all health characteristic dimensions to obtain the composite health index of the target heat exchanger.

[0148] In step 601 of some embodiments, first, according to the dimension weight and dimension correlation coefficient corresponding to each health characteristic dimension, calculate the weighted correlation coefficient corresponding to the health characteristic dimension. The dimension weight reflects the importance of each health characteristic dimension in the composite health index, while the dimension correlation coefficient reflects the degree of proximity between the actual data of each health characteristic dimension and the ideal state. By multiplying the correlation coefficient of each health characteristic dimension by its corresponding weight, the weighted correlation coefficient of this dimension can be obtained. Then, add up the weighted correlation coefficients of all health characteristic dimensions to obtain the final weighted correlation coefficient.

[0149] In step 602 of some embodiments, perform a weighted combination of the weighted correlation coefficients corresponding to all health characteristic dimensions to obtain the composite health index of the target heat exchanger. The composite health index is obtained by weighted combination of multiple single health characteristics and can comprehensively reflect the health state of the heat exchanger. By performing a weighted combination of the weighted correlation coefficients of all health characteristic dimensions, a comprehensive health index can be obtained, which can comprehensively consider multiple degradation characteristics, avoid the limitations of a single index, and improve the accuracy of life prediction.

[0150] In some embodiments, the dimension weights can be adjusted according to the dimension correlation coefficients. For example, if the correlation coefficient of a certain health feature dimension is low, it indicates that the gap between the actual data of this dimension and the ideal state is large, and it may be necessary to reduce the weight of this dimension; conversely, if the correlation coefficient of a certain health feature dimension is high, it indicates that the gap between the actual data of this dimension and the ideal state is small, and it may be necessary to increase the weight of this dimension. In this way, it can be ensured that the weights can more accurately reflect the importance of each health feature dimension. The dimension correlation coefficient reflects the degree of proximity between the actual data of each health feature dimension and the ideal state. If the correlation coefficient of a certain dimension is low, it indicates that the data of this dimension has a large gap from the ideal state, and its weight may need to be reduced; conversely, if the correlation coefficient of a certain dimension is high, it indicates that the data of this dimension is close to the ideal state, and its weight may need to be increased. By dynamically adjusting the weights, it can be ensured that the composite health index can more accurately reflect the health state of the heat exchanger, thereby improving the accuracy and reliability of life prediction.

[0151] In step 305 of some embodiments, the failure threshold is a key indicator for determining whether the heat exchanger has reached the failure state. By analyzing the composite health index data of similar heat exchangers at the time of failure, a reasonable failure threshold can be determined. When the composite health index of the target heat exchanger reaches or exceeds this failure threshold, it indicates that the heat exchanger has approached or reached the failure state and maintenance or replacement is required.

[0152] In step 204 of some embodiments, the model is usually based on the Wiener degradation process model and combines the Kalman filtering algorithm to be able to dynamically predict the degradation process of the heat exchanger. The Wiener degradation process model is used to describe the degradation trend of the equipment, while the Kalman filtering algorithm is used to estimate and update the equipment state in real time. The life prediction model is the core of the method of the present invention and can predict the remaining life of the heat exchanger based on the current equipment state and historical data, providing a scientific basis for equipment maintenance.

[0153] Please refer to Figure 7 , in some embodiments, step 204 may include, but is not limited to, the following steps 701 to 703:

[0154] Step 701, construct a Wiener degradation process model according to the composite health index.

[0155] Step 702, construct a remaining life expression based on the Wiener degradation process model and the failure threshold; wherein, the remaining life expression is used to represent the time required for the target heat exchanger to reach the failure threshold from the current state.

[0156] Step 703, construct a probability density function of the remaining life according to the remaining life expression to obtain the life prediction model corresponding to the target heat exchanger.

[0157] In step 701 of some embodiments, the Wiener degradation process model is a commonly used stochastic process model for describing the degradation process of a device. This model assumes that the degradation process of the device is a stochastic process that changes over time, where the change in the degradation state consists of a deterministic part and a stochastic part. The deterministic part represents the average degradation trend of the device, while the stochastic part represents the uncertainty in the degradation process. By introducing the composite health index into the Wiener degradation process model, the degradation process of the heat exchanger can be described more accurately. The Wiener degradation process model can be expressed as:

[0158] X(t) = X(0) + μt + σW(t)

[0159] Where: X(t) represents the degradation state at time t, and X(0) represents the initial state; μ is the drift parameter, representing the deterministic part in the degradation process, σ is the diffusion parameter, representing the stochastic part in the degradation process, and W(t) represents the standard Wiener process. By introducing the composite health index into the Wiener degradation process model, the degradation process of the heat exchanger can be described more accurately. Then, a Wiener process model combined with the composite health index H i can be constructed, and H i is the degradation rate (i.e., the drift term) affecting the device

[0160] X(t) = X(0) + (μ0 + β·H i )t + σW(t)

[0161] Where μ0 is the basic drift term, representing the basic degradation rate of the heat exchanger device, and β is the influence coefficient of H i , representing the modulation effect of the composite health index on the drift rate.

[0162] In step 702 of some embodiments, the remaining life expression is used to represent the time required for the target heat exchanger to reach the failure threshold from the current state. The failure threshold is a key indicator for judging whether the device has reached the failure state. When the degradation state of the device reaches or exceeds the failure threshold, the device can be regarded as failed. By constructing the remaining life expression, the remaining life of the target heat exchanger can be predicted, providing a scientific basis for equipment maintenance. Construct an expression for the remaining life L. If the failure threshold is X th , when the degradation state X(t) reaches or exceeds X th , the device can be regarded as failed. The remaining life L can be defined as the time difference from the current time t0 to the time when the degradation state reaches the failure threshold X th , that is:

[0163] L = inf{t > t0: X(t) ≥ X th} - t0

[0164] Substituting the above Wiener process model into this definition, we can obtain:

[0165]

[0166] In step 703 of some embodiments, a probability density function of the remaining life is constructed according to the remaining life expression to obtain a life prediction model corresponding to the target heat exchanger. The probability density function is used to describe the distribution of the remaining life, and through the probability density function, the probability of the device failing at a certain time point can be calculated. By constructing the probability density function of the remaining life, a life prediction model of the target heat exchanger can be obtained. Assuming that the current time is t0 = 0, the expression of the probability density function of the life prediction model can be:

[0167]

[0168] where: ΔX = X th - X(0) represents the gap between the initial state and the failure threshold.

[0169] Through steps 701 to 703, by constructing a Wiener degradation process model, a remaining life expression, and a probability density function of the remaining life, a life prediction model of the target heat exchanger can be constructed. The Wiener degradation process model is used to describe the degradation process of the device, the remaining life expression is used to represent the time required for the device to reach the failure threshold from the current state, and the probability density function of the remaining life is used to describe the distribution of the remaining life of the device. These steps significantly improve the accuracy and reliability of life prediction.

[0170] In step 102 of some embodiments, the target measurement point is configured on the target heat exchanger to collect the device data of the target heat exchanger in real time. These data include parameters related to the performance of the heat exchanger such as temperature, pressure, and flow rate. The selection and configuration of the target measurement point are the keys to ensuring the accuracy and comprehensiveness of data collection. The device data collected through the target measurement point can reflect the operating state of the heat exchanger in real time and provide the latest input data for the life prediction model. The target device data is the input of the life prediction model. By collecting the device data in real time, the state of the model can be dynamically updated, improving the accuracy and timeliness of the prediction. The selection of the target measurement point is usually based on the key parts of the device and the optimization of the sensor layout to ensure that the collected data can comprehensively reflect the health state of the device.

[0171] Please refer to Figure 8 , in some embodiments, the process of pre-determining the target measurement point may include, but is not limited to, the following steps 801 to 804:

[0172] Step 801: Based on the historical equipment degradation data, establish a multi-source observation model between the equipment state of the target heat exchanger and the sensor data corresponding to multiple sensor measurement points; wherein, the multi-source observation model is used to describe the mapping relationship between the equipment state and multiple sensor data.

[0173] Step 802: Based on the multi-source observation model, construct a state evaluation coefficient for the target heat exchanger; wherein, the state evaluation coefficient is used to characterize the accuracy of the sensor data corresponding to each sensor measurement point in evaluating the equipment state.

[0174] Step 803: Based on each state evaluation coefficient and a preset resource consumption coefficient, establish a measurement point evaluation function model.

[0175] Step 804: Based on the measurement point evaluation function model, select multiple target measurement points from multiple candidate sensor measurement points.

[0176] In step 801 of some embodiments, the multi-source observation model is used to describe the relationship between the equipment state and the data of multiple sensor measurement points. By establishing this model, the data of multiple sensor measurement points can be associated with the equipment state, thereby providing a basis for subsequent state evaluation and measurement point selection. The equipment state is usually described by a composite health index, while the sensor measurement point data includes parameters such as temperature, pressure, and flow rate. The establishment of the multi-source observation model makes the relationship between the equipment state and the sensor data clearer, providing a basis for subsequent analysis. Among them, the functional relationship between the detected data of a single sensor and the state can describe the observed data as a one-dimensional observation function of the relevant state:

[0177]

[0178] Among them, is the observed value of the p-th measurement point at time k. x k represents the state value of the overall system at time k. is the functional relationship formula describing the change of the observed data with the state, θ p represents its unknown parameters, a p and b p are respectively a scalar parameter and a bias, represents the observation error.

[0179] Considering the correlation between the detected data of multiple sensors, a multi-source observation function between the detected data of multiple sensors and the overall state can be constructed;

[0180]

[0181] Among them, the observation error obeys a multivariate normal distribution.

[0182] In step 802 of some embodiments, the state evaluation coefficient is used to characterize the accuracy of the sensor data corresponding to each sensor measurement point in evaluating the device state. Specifically, the state evaluation coefficient is determined by calculating the error between the observed data of each sensor and the device state. The smaller the error, the more accurate the evaluation of the device state by the observed data of the sensor, and the higher the state evaluation coefficient. The state evaluation error can be expressed as:

[0183]

[0184] where, is the actual observed value of the sensor, is the observed value predicted based on the multi-source observation model. The state evaluation coefficient r a can be expressed as:

[0185]

[0186] where, e p is the state evaluation error, r a is inversely proportional to the error. The smaller the error, the larger r a .

[0187] Please refer to Figure 9 , in some embodiments, step 802 may include, but is not limited to, the following steps 901 to 904:

[0188] Step 901, determine the observation equation of each candidate sensor measurement point according to the multi-source observation model.

[0189] Step 902, construct the state mapping function of the candidate sensor measurement point based on the observation equation.

[0190] Step 903, calculate the state evaluation error corresponding to each candidate sensor measurement point according to the historical device degradation data.

[0191] Step 904, determine the state evaluation coefficient corresponding to each candidate sensor measurement point based on the state evaluation error.

[0192] In step 901 of some embodiments, according to the multi-source observation model, determine the observation equation of each candidate sensor measurement point. The observation equation is used to describe the relationship between the single sensor data and the device state. Specifically, the observation equation is a mathematical function used to map the observed data of a single sensor to the device state. For example, the observation equation of a single sensor can be expressed as:

[0193]

[0194] In step 902 of some embodiments, the state mapping function is used to map sensor data to the device state. Specifically, the state mapping function is a mathematical function that maps the observed data of multiple sensors to the device state. For example, the state mapping function can be expressed as:

[0195] x k = g(Y k ) + ε k

[0196] where represents the set composed of the data of multiple measurement points, and g(·) represents the mapping function relationship from the multi-dimensional observed data Y k to the one-dimensional state data , and ε k represents the state fitting error. Through the state mapping function, the observed data of multiple sensors can be comprehensively mapped to the device state, so as to more comprehensively describe the health state of the device.

[0197] In step 903 of some embodiments, the state evaluation error is used to measure the accuracy of the sensor data for evaluating the device state. Specifically, the state evaluation error is calculated by comparing the difference between the observed data of the sensor and the actual state of the device. For example, the state evaluation error can be expressed as:

[0198]

[0199] where is the actual observed value of the sensor, and is the observed value predicted based on the multi-source observation model.

[0200] In step 904 of some embodiments, the state evaluation coefficient r a can be expressed as:

[0201]

[0202] where e p is the state evaluation error, r a is inversely proportional to the error, and the smaller the error, the larger r a . The state evaluation coefficient reflects the contribution degree of the sensor data to the device state evaluation. The higher the coefficient, the more accurate the observed data of the sensor for the device state evaluation.

[0203] In step 803 of some embodiments, the measurement point evaluation function model is used to evaluate the comprehensive performance of each sensor measurement point. Specifically, the measurement point evaluation function model comprehensively considers the state evaluation coefficient and the resource consumption coefficient. The state evaluation coefficient reflects the accuracy of the sensor data for evaluating the device state, while the resource consumption coefficient reflects the resource consumption such as the power consumption and cost of the sensor.

[0204] Please refer to Figure 10 , in some embodiments, step 803 may include, but is not limited to, the following steps 1001 to 1004:

[0205] Step 1001, construct a selection state space and a selection action space for multiple sensor measurement points; wherein, the selection state space includes state information on whether each sensor measurement point is selected, and the selection action space includes selection operations for each sensor measurement point.

[0206] Step 1002, based on the selection state space and the selection action space, construct a value evaluation function; wherein, the value evaluation function is used to characterize the expected cumulative reward for performing a selection action in the current selection state.

[0207] Step 1003, construct a reward function according to the state evaluation coefficient corresponding to each sensor measurement point and a preset resource consumption coefficient.

[0208] Step 1004, based on the value evaluation function and the reward function, optimize the selection strategy of the sensor measurement points to obtain a measurement point evaluation function model.

[0209] In step 1001 of some embodiments, the selection state space includes state information on whether each sensor measurement point is selected, and the selection action space includes selection operations for each sensor measurement point. Specifically, the selection state space is a binary vector indicating whether each sensor measurement point is selected (1 indicates selected, 0 indicates not selected). For example, if there are N sensors, the selection state space can be expressed as:

[0210] u t = [u1, u2, u N , u i ∈ {0, 1}

[0211] wherein, u j = 1 indicates that the i-th sensor is selected, and u j = 0 indicates that the i-th sensor is not selected. The selection action space is a discrete space indicating the selection operations for each sensor measurement point. For example, the selection action space can be expressed as:

[0212] a t = [u1, u2,..., u N , u i ∈ {0, 1}

[0213] wherein, a t represents the sensor combination selected at time t.

[0214] In step 1002 of some embodiments, the value evaluation function is used to characterize the expected cumulative reward for performing a selection action in the current selection state. Specifically, the value evaluation function is a mathematical function used to evaluate the expected cumulative reward for performing a certain action in the current state. For example, the value evaluation function can be expressed as:

[0215]

[0216] where V(s t ) is the state value function, representing the expected cumulative reward in state s t ; r t is the reward at the current time step; γ is the discount factor, representing the weight of future rewards; s t+1 is the next state after performing action a t . Through the value evaluation function, the long-term benefits of performing a certain action in the current state can be evaluated.

[0217] In step 1003 of some embodiments, the reward function is used to evaluate the comprehensive effect of sensor measurement point selection. Specifically, the reward function comprehensively considers the state evaluation coefficient and the resource consumption coefficient. The state evaluation coefficient reflects the accuracy of sensor data for equipment state evaluation, while the resource consumption coefficient reflects the resource consumption such as the power consumption and cost of the sensor. The reward function can be expressed as:

[0218] r t = w1·r a + w2·r c

[0219] where r a is the state evaluation coefficient, r c is the resource consumption coefficient, and w1 and w2 are weight coefficients used to balance the importance of state evaluation and resource consumption.

[0220] In step 1004 of some embodiments, by optimizing the selection strategy, the sensor measurement point combination that maximizes the cumulative reward is selected. For example, optimization methods such as the greedy algorithm or dynamic programming can be used to select the sensor measurement point combination that maximizes the value of the reward function. The optimization process can be expressed as:

[0221]

[0222] where is the optimal sensor measurement point combination.

[0223] In step 804 of some embodiments, by optimizing the measurement point evaluation function model, a combination of sensor measurement points that maximizes the function value is selected. For example, a reinforcement learning algorithm (such as DQN) is used to select the optimal combination of sensor measurement points. The goal of the reinforcement learning algorithm is to select the optimal combination of sensor measurement points by maximizing the cumulative reward. The policy update formula is as follows:

[0224]

[0225] where α is the learning rate; γ is the discount factor, which represents the weight of future rewards. r t is the reward at the current time step. The policy π(a t |s t ) can be derived through the Q-function, that is, the action that maximizes Q(s, a) is selected:

[0226]

[0227] By continuously updating and repeating the above steps, the optimal measurement point combination can be selected.

[0228] Through steps 801 to 804, the target measurement points can be determined in advance. First, a multi-source observation model is established based on historical equipment degradation data to describe the relationship between the equipment state and the data of multiple sensor measurement points. Then, a state evaluation coefficient is constructed based on the multi-source observation model to evaluate the contribution degree of each sensor measurement point data to the equipment state evaluation. Next, a measurement point evaluation function model is established based on the state evaluation coefficient and the resource consumption coefficient to comprehensively consider the comprehensive performance of each sensor measurement point. Finally, multiple target measurement points are selected from multiple candidate sensor measurement points based on the measurement point evaluation function model to ensure that the selected sensor measurement points can comprehensively and accurately reflect the health state of the equipment while minimizing resource consumption. These steps provide a scientific basis for equipment monitoring, significantly improving the efficiency and accuracy of equipment monitoring.

[0229] In step 103 of some embodiments, the equipment data collected at the target measurement points is input into a pre-constructed life prediction model, and the model will calculate the remaining life of the target heat exchanger based on the current equipment state and historical data. The life prediction model is based on composite health indicators and failure thresholds, and combines the Wiener degradation process model and the Kalman filter algorithm to dynamically predict the degradation process of the heat exchanger. The Kalman filter algorithm can effectively reduce the interference of measurement noise and improve the accuracy of prediction through iterative updates of the equipment state. Through the life prediction model, the health state of the heat exchanger can be evaluated in real time, and its remaining life can be predicted, providing a scientific basis for the equipment maintenance of nuclear power plants and avoiding the occurrence of over-maintenance or under-maintenance situations. The results of life prediction can help nuclear power plants formulate reasonable maintenance plans, reduce equipment maintenance costs, and improve the reliability and safety of equipment operation.

[0230] Please refer to Figure 11 , in some embodiments, step 103 may include, but is not limited to, the following steps 1101 to 1102:

[0231] Step 1101, perform state estimation on the target device data based on the Kalman filter algorithm to obtain a state estimation result.

[0232] Step 1102, perform remaining life prediction on the state estimation result through the life prediction model to obtain the target remaining life of the target heat exchanger.

[0233] In step 1101 of some embodiments, state estimation is performed on the target device data based on the Kalman filter algorithm to obtain a state estimation result. The Kalman filter algorithm is a recursive state estimation algorithm that can effectively process observed data with noise and estimate the state of the system. Specifically, the Kalman filter algorithm includes two main steps: prediction and update.

[0234] In the prediction step, the algorithm predicts the state and covariance matrix at the current moment based on the state equation of the system. The state equation describes how the system state changes over time, while the covariance matrix reflects the uncertainty of the state estimation. Through the prediction step, the algorithm can calculate the state prediction value at the current moment based on the state estimation value at the previous moment.

[0235] In the update step, the algorithm corrects the prediction value based on the actual observed data. The observed data is the device operation data collected by the sensor and usually contains a certain amount of noise. The Kalman filter algorithm calculates the Kalman gain and performs weighted fusion of the observed data and the prediction value to obtain a more accurate state estimation value. Finally, the algorithm outputs the state estimation result at the current moment, which reflects the current health state of the device.

[0236] In step 1102 of some embodiments, the life prediction model first calculates the gap between the current health index of the device and the failure threshold according to the state estimation result. Then, in combination with the degradation rate of the device (including the basic degradation rate and the modulation effect of the composite health index on the degradation rate), the model calculates the remaining life of the device. For example, if the current health index of the device is close to the failure threshold and the degradation rate is fast, the remaining life is short; conversely, if the health index is far from the failure threshold and the degradation rate is slow, the remaining life is long.

[0237] Through the above steps 101 to 103, the method first obtains the historical equipment degradation data of similar heat exchangers and extracts health features from these data to obtain multiple health feature dimensions, thus making full use of the existing operation experience and overcoming the limitations of relying only on empirical formulas. On this basis, the present application constructs a composite health index and its failure threshold based on the extracted multiple health feature dimensions. This composite health index can not only more comprehensively reflect the health state of the equipment, but also reduce the misjudgment risk caused by a single feature by considering the combined effect of multi-dimensional features. Subsequently, the present application combines the composite health index and the failure threshold to construct a dedicated life prediction model, realizing the transformation from static empirical judgment to dynamic data-driven. Finally, by configuring measurement points on the target heat exchanger to collect real-time data and inputting these data into the pre-constructed prediction model, the dynamic prediction of the remaining life of the equipment is realized, and the change trend of the equipment state is captured in a timely manner. The organic combination of the above technical means overcomes the problems of over-maintenance or under-maintenance that may occur in the prior art, significantly improves the accuracy of the life prediction of the nuclear power plant heat exchanger, provides more reliable data support for equipment maintenance decision-making, and further ensures the safe and stable operation of the nuclear power plant.

[0238] Please refer to Figure 12 , the embodiment of the present application also provides a nuclear power plant heat exchanger life prediction device 1200, which can implement the above nuclear power plant heat exchanger life prediction method, including:

[0239] The first acquisition module 1201 is used to acquire a pre-constructed life prediction model; wherein, the life prediction model is pre-constructed by the following steps:

[0240] Obtain the historical equipment degradation data of the similar heat exchangers corresponding to the target heat exchanger;

[0241] Extract health features from the historical equipment degradation data to obtain multiple health feature dimensions;

[0242] Based on multiple health feature dimensions, construct a composite health index and the failure threshold of the composite health index;

[0243] Based on the composite health index and the failure threshold, construct the life prediction model corresponding to the target heat exchanger;

[0244] The second acquisition module 1202 is used to acquire multiple target device data collected by a target measurement point; wherein, the target measurement point is configured on the target heat exchanger and is used to collect the device data of the target heat exchanger;

[0245] The input module 1203 is used to input the multiple target device data into the life prediction model to perform life prediction through the life prediction model and obtain the target remaining life of the target heat exchanger.

[0246] The method for predicting the service life of a heat exchanger in a nuclear power plant according to an embodiment of the present application includes: obtaining a pre-constructed service life prediction model; wherein, the service life prediction model is pre-constructed by the following steps: obtaining historical equipment degradation data of the same type of heat exchanger corresponding to the target heat exchanger; extracting health features from the historical equipment degradation data to obtain multiple health feature dimensions; constructing a composite health index and a failure threshold of the composite health index based on the multiple health feature dimensions; constructing the service life prediction model corresponding to the target heat exchanger based on the composite health index and the failure threshold; obtaining multiple target equipment data collected by a target measurement point; wherein, the target measurement point is configured on the target heat exchanger for collecting equipment data of the target heat exchanger; inputting the multiple target equipment data into the service life prediction model to perform service life prediction through the service life prediction model, so as to obtain the target remaining service life of the target heat exchanger.

[0247] Aiming at the technical problem that the service life prediction of heat exchangers in nuclear power plants in the background art overly relies on empirical formulas and statistical data, the present application provides a method for predicting the service life of heat exchangers based on a composite health index. This method first obtains historical equipment degradation data of the same type of heat exchanger and extracts health features from these data to obtain multiple health feature dimensions, thereby making full use of existing operating experience and overcoming the limitation of relying only on empirical formulas. On this basis, the present application constructs a composite health index and its failure threshold based on the extracted multiple health feature dimensions. This composite health index can not only more comprehensively reflect the health state of the equipment, but also reduce the misjudgment risk caused by a single feature by considering the combined effect of multi-dimensional features. Subsequently, the present application combines the composite health index and the failure threshold to construct a special service life prediction model, realizing the transformation from static empirical judgment to dynamic data-driven. Finally, by configuring measurement points on the target heat exchanger to collect real-time data and inputting these data into the pre-constructed prediction model, the dynamic prediction of the remaining service life of the equipment is realized, and the change trend of the equipment state is captured in a timely manner. The organic combination of the above technical means overcomes the problems of excessive maintenance or insufficient maintenance that may occur in the prior art, significantly improves the accuracy of the service life prediction of heat exchangers in nuclear power plants, provides more reliable data support for equipment maintenance decisions, and further ensures the safe and stable operation of nuclear power plants.

[0248] Refer to Figure 13 , Figure 13 schematically shows the hardware structure of an electronic device in another embodiment. The electronic device includes:

[0249] The processor 1301 can be implemented in the form of a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, etc., and is used to execute relevant programs to implement the technical solutions provided in the embodiments of the present application;

[0250] The memory 1302 can be implemented in the form of a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM), etc. The memory 1302 can store an operating system and other application programs. When implementing the technical solutions provided in the embodiments of this specification through software or firmware, the relevant program codes are stored in the memory 1302 and are called by the processor 1301 to execute the nuclear power plant heat exchanger life prediction method of the embodiments of the present application;

[0251] The input / output interface 1303 is used to implement information input and output;

[0252] The communication interface 1304 is used to implement communication interaction between this device and other devices, and can achieve communication through wired means (such as USB, network cable, etc.) or wireless means (such as mobile network, Wi-Fi, Bluetooth, etc.);

[0253] The bus 1305 transmits information between various components of the device (such as the processor 1301, the memory 1302, the input / output interface 1303, and the communication interface 1304);

[0254] Among them, the processor 1301, the memory 1302, the input / output interface 1303, and the communication interface 1304 achieve communication connections with each other inside the device through the bus 1305.

[0255] The embodiments of the present application also provide a computer program product, which includes a computer program. The processor of the computer device reads and executes this computer program, so that the computer device executes to implement the above-mentioned nuclear power plant heat exchanger life prediction method.

[0256] In the description of the present disclosure and the above-mentioned accompanying drawings, terms such as "first", "second", "third", "fourth", etc. (if any) are used to distinguish similar objects and do not necessarily describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances so that the embodiments of the present disclosure described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "comprising" and "including" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that comprises a series of steps or units does not necessarily limit to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products, or devices.

[0257] It should be understood that in the present disclosure, "at least one (item)" means one or more, and "a plurality" means two or more. "And / or" is used to describe the association relationship of associated objects and indicates that three relationships can exist. For example, "A and / or B" can mean: only A exists, only B exists, and both A and B exist at the same time. Among them, A and B can be singular or plural. The character " / " generally indicates that the associated objects before and after are in an "or" relationship. "At least one (one) of the following" or its similar expression means any combination of these items, including any combination of single item (one) or plural items (ones). For example, at least one (one) of a, b, or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or multiple.

[0258] It should be understood that in the description of the embodiments of the present application, the meaning of "a plurality (or multiple items)" is more than two. Understandings such as "greater than", "less than", "exceeding", etc. do not include the present number, and understandings such as "above", "below", "within", etc. include the present number.

[0259] In several embodiments provided by the present disclosure, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. For example, the device embodiments described above are only illustrative. For example, the division of units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed coupling or direct coupling or communication connection to each other can be through some interfaces, and the indirect coupling or communication connection of devices or units can be in electrical, mechanical, or other forms.

[0260] The unit described as a separation component may or may not be physically separated. The component presented as a unit may or may not be a physical unit, that is, it may be located in one place or distributed across multiple network units. Some or all of these units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0261] In addition, in each embodiment of the present disclosure, each functional unit can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of a software functional unit.

[0262] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present disclosure, in essence, or the part that contributes to the prior art, or all or part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods in each embodiment of the present disclosure. The aforementioned storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical discs that can store program codes.

[0263] It should also be understood that the various embodiments provided in the embodiments of this application can be combined arbitrarily to achieve different technical effects.

[0264] The above is a specific description of the embodiments of the present disclosure, but the present disclosure is not limited to the above embodiments. Those skilled in the art can also make various equivalent deformations or substitutions without departing from the spirit of the present disclosure, and these equivalent deformations or substitutions are all included within the scope defined by the claims of the present disclosure.

Claims

1. A method for predicting the life of a heat exchanger in a nuclear power plant, characterized in that: The method comprises: Obtain a pre-built life prediction model; wherein the life prediction model is pre-built by the following steps: Acquire historical equipment degradation data of similar heat exchangers corresponding to the target heat exchanger; Extracting health features from the historical equipment degradation data to obtain multiple health feature dimensions; Based on the multiple health characteristic dimensions, construct a composite health indicator and a failure threshold of the composite health indicator; Based on the composite health index and the failure threshold, constructing the life prediction model corresponding to the target heat exchanger; Acquire multiple target device data collected by the target measuring point; wherein the target measuring point is configured at the target heat exchanger and is used to collect the device data of the target heat exchanger; The plurality of target equipment data are input into the life prediction model to perform life prediction through the life prediction model to obtain the target remaining life of the target heat exchanger.

2. The method for predicting the life of a heat exchanger in a nuclear power plant according to claim 1, characterized in that: Each of the health feature dimensions includes a plurality of sample data, and the construction of a composite health indicator and a failure threshold of the composite health indicator based on the plurality of health feature dimensions includes: Constructing a health indicator decision matrix according to the multiple health characteristic dimensions and the multiple sample data; Based on the health indicator decision matrix, calculating the dimension weight corresponding to each health characteristic dimension; Calculate the dimension correlation coefficient corresponding to each health feature dimension according to the preset multi-dimensional health feature reference sequence; Constructing a composite health index of the target heat exchanger according to the dimension weight and the correlation coefficient corresponding to each health feature dimension; Based on the historical equipment degradation data of the same type of heat exchanger when it fails, a failure threshold of the composite health indicator is determined.

3. The method for predicting the life of a heat exchanger in a nuclear power plant according to claim 2, characterized in that: The calculating the dimension weight corresponding to each health characteristic dimension based on the health indicator decision matrix includes: Calculating the feature probability corresponding to each of the health feature dimensions according to the health indicator decision matrix; Calculate the dimension entropy value corresponding to each health feature dimension according to the feature probability corresponding to each health feature dimension; According to the dimensional entropy value corresponding to each of the health characteristic dimensions, the dimensional weight corresponding to the health characteristic dimension is calculated.

4. The method for predicting the life of a heat exchanger in a nuclear power plant according to claim 2, characterized in that: The step of calculating the dimension correlation coefficient corresponding to each health feature dimension according to the preset multi-dimensional health feature reference sequence includes: Calculate the difference between each sample data and the corresponding data of the reference sequence in each health feature dimension; determining a maximum difference value and a minimum difference value among all said difference values; The dimension association coefficient corresponding to each of the health feature dimensions is determined according to the difference value, the maximum difference value, the minimum difference value and a preset resolution coefficient.

5. The method for predicting the life of a heat exchanger in a nuclear power plant according to claim 2, characterized in that: The step of constructing a composite health index of the target heat exchanger according to the dimension weight and the correlation coefficient corresponding to each health feature dimension includes: Calculate the weighted correlation coefficient corresponding to each health feature dimension according to the dimension weight and dimension correlation coefficient corresponding to each health feature dimension; The weighted correlation coefficients corresponding to all the health feature dimensions are weightedly combined to obtain a composite health index of the target heat exchanger.

6. The method for predicting the life of a heat exchanger in a nuclear power plant according to claim 1, characterized in that: The constructing the life prediction model corresponding to the target heat exchanger based on the composite health index and the failure threshold includes: constructing a Weiner degradation process model based on the composite health indicator; Based on the Wiener degradation process model and the failure threshold, a remaining life expression is constructed; wherein the remaining life expression is used to represent the time required for the target heat exchanger to reach the failure threshold from the current state; A probability density function of the remaining life is constructed according to the remaining life expression to obtain a life prediction model corresponding to the target heat exchanger.

7. The method for predicting the life of a heat exchanger in a nuclear power plant according to claim 1, characterized in that: The process of predetermining the target measuring point includes: Based on the historical equipment degradation data, a multi-source observation model between the equipment state of the target heat exchanger and the sensor data corresponding to the multiple sensor measurement points is established; wherein the multi-source observation model is used to describe the mapping relationship between the equipment state and the multiple sensor data; Based on the multi-source observation model, construct a state evaluation coefficient of the target heat exchanger; wherein the state evaluation coefficient is used to characterize the accuracy of the sensor data corresponding to each of the sensor measurement points in evaluating the equipment state; Based on each of the state evaluation coefficients and the preset resource consumption coefficient, a measuring point evaluation function model is established; wherein the measuring point evaluation function model is used to evaluate the comprehensive performance of each sensor measuring point; Based on the measuring point evaluation function model, a plurality of target measuring points are selected from a plurality of candidate sensor measuring points.

8. The method for predicting the life of a heat exchanger in a nuclear power plant according to claim 7, characterized in that: The step of constructing a state assessment coefficient of the target heat exchanger based on the multi-source observation model includes: According to the multi-source observation model, an observation equation for each candidate sensor measurement point is determined; wherein the observation equation is used to describe the relationship between a single sensor data and a device state; Based on the observation equation, construct a state mapping function for each candidate sensor measurement point; wherein the state mapping function is used to map sensor data to device state; According to the historical equipment degradation data, the state assessment error corresponding to each candidate sensor measurement point is calculated; Based on the state evaluation error, a state evaluation coefficient corresponding to each candidate sensor measuring point is determined.

9. The method for predicting the life of a heat exchanger in a nuclear power plant according to claim 7, characterized in that: The step of establishing a measuring point evaluation function model based on each of the state evaluation coefficients and the preset resource consumption coefficient comprises: Constructing a selection state space and a selection action space of the plurality of sensor measurement points; wherein the selection state space includes state information of whether each of the sensor measurement points is selected, and the selection action space includes selection operations of each of the sensor measurement points; Based on the selection state space and the selection action space, construct a value evaluation function; wherein the value evaluation function is used to characterize the expected cumulative reward of performing the selection action in the current selection state; Constructing a reward function according to the state evaluation coefficient corresponding to each of the sensor measurement points and a preset resource consumption coefficient; Based on the value assessment function and the reward function, the selection strategy of the sensor measuring points is optimized to obtain the measuring point evaluation function model.

10. The method for predicting the life of a heat exchanger in a nuclear power plant according to claim 1, characterized in that: The performing life prediction by using the life prediction model to obtain the target remaining life of the target heat exchanger includes: Performing state estimation on the target device data based on a Kalman filter algorithm to obtain a state estimation result; The life prediction model is used to perform life prediction on the state estimation result to obtain the target remaining life of the target heat exchanger.

11. A device for predicting the life of a heat exchanger in a nuclear power plant, characterized in that: include: The first acquisition module is used to acquire a pre-built life prediction model; wherein the life prediction model is pre-built by the following steps: Acquire historical equipment degradation data of similar heat exchangers corresponding to the target heat exchanger; Extracting health features from the historical equipment degradation data to obtain multiple health feature dimensions; Based on the multiple health characteristic dimensions, construct a composite health indicator and a failure threshold of the composite health indicator; Based on the composite health index and the failure threshold, constructing the life prediction model corresponding to the target heat exchanger; A second acquisition module is used to acquire a plurality of target device data collected by a target measuring point; wherein the target measuring point is configured at a target heat exchanger and is used to collect device data of the target heat exchanger; An input module is used to input the plurality of target device data into the life prediction model, so as to perform life prediction through the life prediction model to obtain the target remaining life of the target heat exchanger.

12. An electronic device, characterized in that: include: A memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the method for predicting the life of a heat exchanger in a nuclear power plant as claimed in any one of claims 1 to 10 is implemented.

13. A computer-readable storage medium, characterized in that: The storage medium stores a program, and the program is executed by a processor to implement the method for predicting the life of a heat exchanger in a nuclear power plant as described in any one of claims 1 to 10.

Images