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

Abstract

The invention relates to the technical field of heat exchanger fault detection, in particular to a heat exchanger fault diagnosis and prediction maintenance system and method based on data analysis. Working data of the heat exchanger are preprocessed, a qualified interval of the heat exchanger is set according to the preprocessed working data, then a heat exchanger standard model is created and trained, the trained heat exchanger standard model is obtained, and finally whether the qualification rate of the heat exchanger is larger than or equal to a qualification rate threshold value or not is judged based on real-time evaluation of the heat exchanger. If the qualified rate of the heat exchanger is larger than or equal to the qualified rate threshold value, a suspected area is obtained, the qualified rate trend of the heat exchanger is calculated in combination with the suspected area and the unqualified area, and whether early warning needs to be sent out or not is judged based on the qualified rate trend of the heat exchanger. And the qualified rate trend of the heat exchanger is calculated in combination with the suspected area and the unqualified area, so that early warning is given out in time before the heat exchanger is damaged.

Description

Technical Field

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

[0002] Heat exchanger is an energy-saving device that transfers heat between two or more fluids at different temperatures. It transfers heat from a higher temperature fluid to a lower temperature fluid, so that the fluid temperature reaches the specified index of the process to meet the needs of process conditions. It is also one of the main equipment for improving energy utilization. The heat exchanger industry involves nearly 30 industries such as HVAC, pressure vessels, reclaimed water treatment equipment, chemicals, petroleum, etc., forming an industrial chain.

[0003] A Chinese patent with publication number CN118067422A discloses a method, device, equipment and medium for diagnosing faults of a heat exchanger. The method calculates the statistics of the operating data to be detected based on the operating data to be detected and the standard distance k-nearest neighbor algorithm; determines whether the statistics of the operating data to be detected is greater than the control upper limit of the confidence level, and determines whether the heat exchanger is faulty based on the judgment result. However, in the prior art, the current state of the heat exchanger is judged based on real-time data, but the state trend of the heat exchanger is not predicted, which makes it difficult to issue a warning in time before an accident occurs. Summary of the invention

[0004] The purpose of the present invention is to address the problems existing in the background technology and to propose a heat exchanger fault diagnosis and predictive maintenance system and method based on data analysis.

[0005] The technical solution of the present invention: On the one hand, the present application provides a heat exchanger fault diagnosis and predictive maintenance method based on data analysis, including: Setting acquisition parameters, acquiring working data of multiple heat exchangers based on the acquisition parameters, preprocessing the working data of the heat exchangers to obtain preprocessed working data, and setting a qualified interval of the heat exchanger according to the preprocessed working data; Creating a heat exchanger standard model, and training the heat exchanger standard model by preprocessing the working data, so that the heat exchanger standard model determines the working state of the heat exchanger based on the working data of the heat exchanger, and obtaining the trained heat exchanger standard model; Collecting real-time data of the heat exchanger, inputting the real-time data of the heat exchanger into the trained standard model of the heat exchanger to obtain a real-time evaluation of the heat exchanger, and judging whether the qualified rate of the heat exchanger is greater than or equal to the qualified rate threshold based on the real-time evaluation of the heat exchanger; If the qualified rate of the heat exchanger is greater than or equal to the qualified rate threshold, the suspected area is obtained and the qualified rate trend of the heat exchanger is calculated by combining the suspected area with the unqualified area, and whether an early warning needs to be issued is determined based on the qualified rate trend of the heat exchanger.

[0006] Preferably, setting acquisition parameters, acquiring working data of a plurality of heat exchangers based on the acquisition parameters, preprocessing the working data of the heat exchangers to obtain preprocessed working data, and setting a qualified interval of the heat exchanger according to the preprocessed working data, including: Create a heat exchanger data table; Divide the heat exchanger into multiple heat exchange areas; The acquisition parameters are respectively set for the multiple heat exchange areas of each heat exchanger; the acquisition parameters include acquisition period, acquisition frequency and acquisition position; For each heat exchanger, the working data of multiple heat exchange areas of the heat exchanger are collected according to the collection parameters of the heat exchanger, and the collected working data are put into the heat exchanger data table; the working data includes multiple collection items of heat exchanger temperature data, heat exchanger pressure data and heat exchanger flow data.

[0007] Preferably, setting acquisition parameters, acquiring working data of a plurality of heat exchangers based on the acquisition parameters, preprocessing the working data of the heat exchangers to obtain preprocessed working data, and setting a qualified interval of the heat exchanger according to the preprocessed working data, further comprising: Randomly select a heat exchanger from the heat exchanger data table and obtain the working data of the heat exchanger; Select a collection item and calculate the average value of the collection item; For this collection project, determine whether there is working data under each collection node in order according to the order of collection time; If there is no working data under the collection node, the average value of the collection item is filled into the collection node to complete the working data; Set a deviation threshold, calculate the upper and lower boundary values ​​based on the mean calculation of the collection item and the deviation threshold, and set the qualified interval based on the upper and lower boundary values; Return to select a collection item and calculate the average value of the collection item until the qualified interval of each collection item of the heat exchanger is obtained; Return to randomly select a heat exchanger from the heat exchanger data table until all heat exchangers in the working data table are selected, and obtain the qualified interval of each collection item of all heat exchangers.

[0008] Preferably, a heat exchanger standard model is created, and the heat exchanger standard model is trained by preprocessing working data so that the heat exchanger standard model determines the working state of the heat exchanger based on the working data of the heat exchanger, and the trained heat exchanger standard model is obtained, including: Create a standard model of a heat exchanger; Divide all the pre-processed working data of the heat exchanger into training set and test set in random proportion; The training set is input into the heat exchanger standard model, so that the heat exchanger standard model continuously learns the coupling relationship between the heat exchanger working data and the heat exchanger state, and obtains the trained heat exchanger standard model; The test set is input into the trained heat exchanger standard model to verify whether the trained heat exchanger standard model is trained.

[0009] Preferably, the training set is input into the heat exchanger standard model so that the heat exchanger standard model continuously learns the coupling relationship between the heat exchanger working data and the heat exchanger state, and obtains the trained heat exchanger standard model, including: For each heat exchanger in the training set, obtain the working data of each heat exchanger respectively; Select a heat exchanger and determine in turn whether the working data of each collected item of the heat exchanger is within the qualified range; If the working data of the collected items is within the qualified range, the heat exchanger is recorded as qualified; If the working data of the collected items is not within the qualified range, the heat exchanger will be recorded as unqualified; Return to select a heat exchanger until all heat exchangers are selected and the working status of each heat exchanger is obtained; Based on the working data of each heat exchanger and the working state under the working data, a coupling relationship of heat exchanger-working data-working state is established, and the coupling relationship of heat exchanger-working data-working state is used as a training sample, thereby obtaining multiple training samples; Multiple training samples are sequentially input into the heat exchanger standard model, so that the heat exchanger standard model continuously learns the corresponding relationship between the heat exchanger working data and the heat exchanger state, and obtains the trained heat exchanger standard model.

[0010] Preferably, real-time data of the heat exchanger is collected, and the real-time data of the heat exchanger is input into a trained standard model of the heat exchanger to obtain a real-time evaluation of the heat exchanger, and whether the qualified rate of the heat exchanger is greater than or equal to a qualified rate threshold is judged based on the real-time evaluation of the heat exchanger, including: Collect real-time data of each heat exchange area of ​​the heat exchanger; Input the real-time data of each heat exchange area of ​​the heat exchanger into the trained standard model of the heat exchanger, and obtain the real-time evaluation of each heat exchange area of ​​the heat exchanger output by the trained standard model of the heat exchanger; Set pass rate threshold; Determine in turn whether the real-time evaluation of each heat exchange area of ​​the heat exchanger is qualified, and calculate the qualified rate of all heat exchange areas of the heat exchanger; Determine whether the qualified rate of all heat exchange areas of the heat exchanger is greater than or equal to the qualified rate threshold; If the qualified rate of a heat exchange area of ​​the heat exchanger is lower than the qualified rate threshold, the heat exchanger is stopped and an early warning is issued.

[0011] Preferably, if the qualified rate of the heat exchanger is greater than or equal to the qualified rate threshold, the suspected area is obtained and the qualified rate trend of the heat exchanger is calculated by combining the suspected area with the unqualified area, and whether an early warning needs to be issued is determined based on the qualified rate trend of the heat exchanger, including: If the qualified rate of all heat exchange areas of the heat exchanger is greater than or equal to the qualified rate threshold, it is determined whether there is an unqualified heat exchange area; If there is no unqualified heat exchange area, keep the heat exchanger working; If there is an unqualified heat exchange area, a suspected area is obtained according to the unqualified heat exchange area.

[0012] Preferably, if there is an unqualified heat exchange area, obtaining a suspected area according to the unqualified heat exchange area includes: Select an unqualified heat exchange area; Obtaining the adjacent heat exchange regions of the unqualified heat exchange region; Determine whether the working status of adjacent heat exchange areas is qualified; If the working state of the adjacent heat exchange area is qualified, the adjacent heat exchange area is recorded as an observation area, and the historical working data of the observation area is obtained; Set the error threshold; Calculate the error between the historical working data of the observation area and the historical working data of the corresponding unqualified heat exchange area, and determine whether the error between the historical working data of the observation area and the historical working data of the corresponding unqualified heat exchange area is greater than or equal to an error threshold; If the error between the historical working data of the observed area and the historical working data of the corresponding unqualified heat exchange area is less than or equal to the error threshold, the observed area is recorded as a suspected area; Return to select an unqualified heat exchange area until all unqualified heat exchange areas are selected, and multiple suspected areas are obtained.

[0013] Preferably, if the qualified rate of the heat exchanger is greater than or equal to the qualified rate threshold, the suspected area is obtained and the qualified rate trend of the heat exchanger is calculated by combining the suspected area with the unqualified area, and whether an early warning needs to be issued is determined based on the qualified rate trend of the heat exchanger, which also includes: The qualified rate of the heat exchanger is recalculated by combining the suspected area with the unqualified heat exchange area to obtain the trend qualified rate of the heat exchanger; Determine whether the trend qualification rate of the heat exchanger is greater than or equal to the qualification rate threshold; If the trend qualified rate of the heat exchanger is greater than or equal to the qualified rate threshold, the operation of the heat exchanger is maintained; If the trend pass rate of the heat exchanger is less than the pass rate threshold, an early warning is issued.

[0014] On the other hand, the present application also provides a heat exchanger fault diagnosis and predictive maintenance system based on data analysis, including: A collection unit, which collects working data of the heat exchanger through the collection component; A data processing unit, through which any of the aforementioned methods for heat exchanger fault diagnosis and predictive maintenance based on data analysis is executed, the data processing unit includes a data processing subunit and a control subunit, the working data of the heat exchanger is processed through the data processing subunit, and the working state of the heat exchanger is controlled through the control subunit.

[0015] Compared with the prior art, the above technical solution of the present invention has the following beneficial technical effects: Working data of multiple heat exchangers are collected and preprocessed to obtain preprocessed working data. A qualified interval of the heat exchanger is set according to the preprocessed working data. A standard model of the heat exchanger is then created and trained, so that the standard model of the heat exchanger judges the working state of the heat exchanger based on the working data of the heat exchanger to obtain a trained standard model of the heat exchanger. The real-time data of the heat exchanger is input into the trained standard model of the heat exchanger to obtain a real-time evaluation of the heat exchanger. Based on the real-time evaluation of the heat exchanger, it is judged whether the qualified rate of the heat exchanger is greater than or equal to the qualified rate threshold. If the qualified rate of the heat exchanger is greater than or equal to the qualified rate threshold, a suspected area is obtained and the qualified rate trend of the heat exchanger is calculated in combination with the suspected area and the unqualified area. Based on the qualified rate trend of the heat exchanger, it is judged whether an early warning needs to be issued. The present application finds the suspected area of ​​the heat exchanger and calculates the qualified rate trend of the heat exchanger in combination with the suspected area and the unqualified area, thereby issuing an early warning in time before the heat exchanger is damaged. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 A schematic flow chart of a heat exchanger fault diagnosis and predictive maintenance method based on data analysis proposed in the present invention; Figure 2 This is a schematic diagram of the structure of the heat exchanger fault diagnosis and predictive maintenance system based on data analysis proposed by the present invention.

[0017] Reference numerals: 100, acquisition unit; 200, data processing unit; 201, data processing subunit; 202. Control subunit. DETAILED DESCRIPTION

[0018] Embodiment 1, as Figure 1 As shown, the heat exchanger fault diagnosis and predictive maintenance method based on data analysis proposed by the present invention includes: S100, setting acquisition parameters, acquiring working data of multiple heat exchangers based on the acquisition parameters, preprocessing the working data of the heat exchangers to obtain preprocessed working data, and setting a qualified interval of the heat exchanger according to the preprocessed working data; S200, creating a heat exchanger standard model, and training the heat exchanger standard model by preprocessing the working data, so that the heat exchanger standard model determines the working state of the heat exchanger based on the working data of the heat exchanger, and obtains the trained heat exchanger standard model; S300, collecting real-time data of the heat exchanger, inputting the real-time data of the heat exchanger into the trained standard model of the heat exchanger, obtaining a real-time evaluation of the heat exchanger, and judging whether the qualified rate of the heat exchanger is greater than or equal to the qualified rate threshold based on the real-time evaluation of the heat exchanger; S400: If the qualified rate of the heat exchanger is greater than or equal to the qualified rate threshold, the suspected area is obtained and the qualified rate trend of the heat exchanger is calculated by combining the suspected area with the unqualified area, and whether an early warning needs to be issued is determined based on the qualified rate trend of the heat exchanger.

[0019] In the present invention, working data of multiple heat exchangers are collected, the working data of the heat exchangers are preprocessed to obtain preprocessed working data, and a qualified interval of the heat exchanger is set according to the preprocessed working data. Then, a standard model of the heat exchanger is created and trained, so that the standard model of the heat exchanger judges the working state of the heat exchanger based on the working data of the heat exchanger to obtain a trained standard model of the heat exchanger. The real-time data of the heat exchanger is input into the trained standard model of the heat exchanger to obtain a real-time evaluation of the heat exchanger. Based on the real-time evaluation of the heat exchanger, it is judged whether the qualified rate of the heat exchanger is greater than or equal to the qualified rate threshold. If the qualified rate of the heat exchanger is greater than or equal to the qualified rate threshold, a suspected area is obtained and the qualified rate trend of the heat exchanger is calculated in combination with the suspected area and the unqualified area. Based on the qualified rate trend of the heat exchanger, it is judged whether an early warning needs to be issued. The present application finds the suspected area of ​​the heat exchanger and calculates the qualified rate trend of the heat exchanger in combination with the suspected area and the unqualified area, thereby issuing an early warning in time before the heat exchanger is damaged.

[0020] In an optional embodiment, the S100 includes: S110, creating a heat exchanger data table; S120, dividing each heat exchanger into a plurality of heat exchange areas; Specifically, for different heat exchange areas of the same heat exchanger, the areas of different heat exchange areas should be equal when dividing the heat exchange areas; S130, respectively setting acquisition parameters for the multiple heat exchange areas of each heat exchanger; the acquisition parameters include an acquisition period, an acquisition frequency, and an acquisition position; S140, for each heat exchanger, collect the working data of multiple heat exchange areas of the heat exchanger according to the collection parameters of the heat exchanger, and put the collected working data into the heat exchanger data table; the working data includes multiple collection items of heat exchanger temperature data, heat exchanger pressure data and heat exchanger flow data.

[0021] It should be noted that by collecting the working data of multiple heat exchangers and taking the working data of multiple heat exchangers as training samples to train the heat exchanger standard model, each heat exchanger is divided into multiple heat exchange areas. This is to ensure that when verifying whether the pass rate of the heat exchanger is greater than or equal to the pass rate threshold, the heat exchanger can be regarded as multiple independent areas, thereby realizing a more detailed control method for the heat exchanger.

[0022] In an optional embodiment, the S100 includes: S150, randomly selecting a heat exchanger from the heat exchanger data table and obtaining the working data of the heat exchanger; S160, selecting a collection item and calculating the average value of the collection item; S170, for the collection project, determine whether there is working data under each collection node in sequence according to the order of collection time; S180, if there is no working data under the collection node, fill the average value of the collection item into the collection node to complete the working data; S190, setting a deviation threshold, calculating upper and lower boundary values ​​by combining the mean calculation of the collection item with the deviation threshold, and setting a qualified interval based on the upper and lower boundary values; S191, return to step S160, until a qualified interval of each acquisition item of the heat exchanger is obtained; S192, returning to step S150, until all heat exchangers in the working data table are selected, and the qualified intervals of each collection item of all heat exchangers are obtained; Optionally, in order to avoid the situation where working data is missing due to damage to the collection unit, a missing threshold can be set, and it can be determined whether the number of missing data of the collection project is greater than or equal to the missing threshold. If the number of missing data of the collection project is greater than or equal to the missing threshold, the working data of the collection project is deleted.

[0023] It should be noted that the qualified interval of the collection items is set based on the average value of each collection item. When the working data of the heat exchanger is within the qualified interval, it means that the heat exchanger is in a normal working state. When the working data of the heat exchanger is not within the qualified interval, it means that the heat exchanger may be abnormal.

[0024] When calculating the qualified interval, if the average value of collection project A is A, and the deviation threshold is set to 20%, then the qualified interval of collection project A is [A*80%, A*120%]. For different collection projects, the deviation threshold is set differently. The smaller the deviation threshold is set, the smaller the range of the qualified interval of the collection project is, and the higher the qualified requirements for the collection project are.

[0025] In an optional embodiment, the step S200 includes: S210, create a standard model of heat exchanger; S220, dividing all pre-processed working data of the heat exchanger into a training set and a test set in a random ratio; S230, inputting the training set into the heat exchanger standard model, so that the heat exchanger standard model continuously learns the coupling relationship between the heat exchanger working data and the heat exchanger state, and obtains a trained heat exchanger standard model; S240, inputting the test set into the trained heat exchanger standard model to verify whether the trained heat exchanger standard model is trained.

[0026] It should be noted that by creating and training a heat exchanger standard model, the trained heat exchanger standard model can determine the working status of the heat exchanger according to the input heat exchanger working data, and determine whether an early warning needs to be issued based on the working status of the heat exchanger. The heat exchanger standard model can perform all-day and all-round detection, thereby improving the detection efficiency of the heat exchanger.

[0027] In an optional embodiment, the S230 includes: S231, for each heat exchanger in the training set, respectively obtain the working data of each heat exchanger; S232, selecting a heat exchanger, and determining in turn whether the working data of each collected item of the heat exchanger is within a qualified range; S233, if the working data of the collected item is within the qualified range, the heat exchanger is recorded as being in a qualified state; S234, if the working data of the collected item is not within the qualified range, the heat exchanger is recorded as unqualified; S235, returning to select a heat exchanger until all heat exchangers are selected, and obtaining the working status of each heat exchanger; S236, establishing a coupling relationship of heat exchange area-working data-working state based on the working data of each heat exchanger and the working state under the working data, and taking the heat exchanger, the working data, and the coupling relationship of heat exchange area-working data-working state as a training sample, thereby obtaining multiple training samples; S237, inputting a plurality of training samples into the heat exchanger standard model in sequence, so that the heat exchanger standard model continuously learns the correspondence between the working data of the heat exchange area of ​​the heat exchanger and the state of the heat exchanger, and obtains the trained heat exchanger standard model.

[0028] It should be noted that when training the heat exchanger standard model, the heat exchanger, working data, and heat exchange area-working data-working status are input into the heat exchanger standard model as a training sample, so that the heat exchanger standard model continuously learns the relationship between the working data and the working status.

[0029] After obtaining the trained heat exchanger standard model, the working data of the heat exchanger in the test set is input into the trained heat exchanger standard model to obtain the predicted working state output by the trained heat exchanger standard model, and whether the predicted working state is consistent with the working state in the test set is checked. If the predicted working state is consistent with the working state in the test set, it proves that the trained heat exchanger standard model has been trained. If there are some predicted working states that are inconsistent with the working states in the test set, return to step S220 until the predicted working state is consistent with the working state in the test set.

[0030] In an optional embodiment, the S300 includes: S310, collecting real-time data of each heat exchange area of ​​the heat exchanger; S320, inputting the real-time data of each heat exchange area of ​​the heat exchanger into the trained heat exchanger standard model, and obtaining the real-time evaluation of each heat exchange area of ​​the heat exchanger output by the trained heat exchanger standard model; Specifically, the real-time evaluation includes whether the heat exchange area is qualified or the heat exchange area is unqualified; S330, setting a pass rate threshold; S340, determining in turn whether the real-time evaluation of each heat exchange area of ​​the heat exchanger is qualified, and calculating the qualified rate of all heat exchange areas of the heat exchanger; S350, determining whether the qualified rate of all heat exchange areas of the heat exchanger is greater than or equal to the qualified rate threshold; S360: If the qualified rate of all heat exchange areas of the heat exchanger is less than the qualified rate threshold, the heat exchanger is stopped and an early warning is issued.

[0031] It should be noted that since the heat exchanger is divided into multiple heat exchange areas, for a heat exchanger, each heat exchange area can be judged separately, so that the normal operation of the heat exchanger can be maintained when an unqualified heat exchange area does not affect the normal operation of the heat exchanger.

[0032] When judging whether the heat exchanger can work normally, a pass rate threshold is set, and it is judged whether the total pass rate of all heat exchange areas included in the heat exchanger is greater than or equal to the pass rate threshold. When the pass rate of all heat exchange areas of the heat exchanger is less than the pass rate threshold, it is necessary to stop the operation of the heat exchanger in time to avoid causing greater damage to the heat exchanger.

[0033] The present application divides the heat exchanger into heat exchange areas. When the proportion of unqualified heat exchange areas is less than the qualified rate threshold, the heat exchanger can be kept operating normally and an early warning can be issued to the user, thereby enabling the user to promptly detect heat exchanger abnormalities while ensuring the heat exchange efficiency of the heat exchanger.

[0034] In an optional embodiment, the S400 includes: S410, if the qualified rate of all heat exchange areas of the heat exchanger is greater than or equal to the qualified rate threshold, determine whether there is an unqualified heat exchange area; S420, if there is no unqualified heat exchange area, keep the heat exchanger in operation; S430, if there is an unqualified heat exchange area, obtaining a suspected area according to the unqualified heat exchange area; Specifically, the suspected area refers to a heat exchange area that is still in a qualified state but may be damaged. For such a heat exchange area, it is necessary to increase the detection intensity so as to issue a warning in time before affecting the normal operation of the heat exchanger.

[0035] It should be noted that when the pass rate of all heat exchange areas of the heat exchanger is greater than or equal to the pass rate threshold, there are still two situations, namely, all heat exchange areas are qualified and there are some unqualified heat exchange areas. For these two situations, the present application still adopts different control strategies. For the situation that all heat exchange areas are qualified, it can represent that the heat exchanger is intact and there is no need to issue an early warning. It is sufficient to maintain the normal working state of the heat exchanger. For the situation that there are some unqualified heat exchange areas, it means that the heat exchanger may be on the verge of damage. Therefore, in this case, subsequent inspection of the heat exchanger is required to issue an early warning.

[0036] If for heat exchanger A, its corresponding heat exchange areas are heat exchange area A-1, heat exchange area A-2, heat exchange area A-3, heat exchange area A-4 and heat exchange area A-5, and the set pass rate threshold is 80%, then after the working data of each heat exchange area is input into the trained heat exchanger standard model respectively, the working states obtained are qualified, qualified, qualified, unqualified and qualified, then for heat exchanger A, its pass rate is 80%, which is equal to the pass rate threshold. In this case, heat exchanger A can continue to work, but it is necessary to find the suspected area of ​​the heat exchanger.

[0037] In an optional embodiment, the S430 includes: S431, selecting an unqualified heat exchange area; S432, obtaining the adjacent heat exchange regions of the unqualified heat exchange region; S433, judging whether the working status of the adjacent heat exchange area is qualified; S434, if the working state of the adjacent heat exchange area is qualified, the adjacent heat exchange area is recorded as an observation area, and the historical working data of the observation area is obtained; S435, setting an error threshold; S436, calculating the error between the historical working data of the observation area and the historical working data of the corresponding unqualified heat exchange area, and determining whether the error between the historical working data of the observation area and the historical working data of the corresponding unqualified heat exchange area is greater than or equal to an error threshold; Specifically, when calculating the error between the historical working data of the observation area and the historical working data of the corresponding unqualified heat exchange area, the working data of the first N collection nodes located at the collection nodes for real-time data collection are calculated, and the calculation needs to be performed for each collection item; S437, if the error between the historical working data of the observed area and the historical working data of the corresponding unqualified heat exchange area is less than or equal to the error threshold, the observed area is recorded as a suspected area; S438, returning to select an unqualified heat exchange area, until all unqualified heat exchange areas are selected, and multiple suspected areas are obtained.

[0038] It should be noted that, in actual use, the heat exchange area near the unqualified heat exchange area is also more prone to damage. Therefore, in the process of looking for suspected areas, the qualified areas adjacent to the unqualified heat exchange area are first observed, and the error value between the historical working data of the observation area and the historical working data of the unqualified heat exchange area is calculated. If the error value is smaller, it means that the working state of the observation area is closer to the working state of the unqualified heat exchange area, and the damage is more likely to occur. Therefore, when the error is less than or equal to the error threshold, it can be considered that the observation area has a greater possibility of damage, and the observation area is recorded as a suspected area.

[0039] In an optional embodiment, the S400 further includes: S440, recalculating the qualified rate of the heat exchanger by combining the suspected area with the unqualified heat exchange area to obtain the trend qualified rate of the heat exchanger; S450, determining whether the trend qualified rate of the heat exchanger is greater than or equal to a qualified rate threshold; S460, if the trend qualified rate of the heat exchanger is greater than or equal to the qualified rate threshold, then maintain the operation of the heat exchanger; S470: If the trend qualification rate of the heat exchanger is less than the qualification rate threshold, an early warning is issued.

[0040] It should be noted that when recalculating the pass rate of the heat exchanger, the suspected area needs to be regarded as an unqualified heat exchange area to play a predictive role. After taking into account the suspected area, the pass rate of the heat exchanger will definitely decrease. If the trend pass rate of the heat exchanger is less than the pass rate threshold, it is considered that the heat exchanger has a trend of not being able to work normally, so an early warning needs to be issued to the user.

[0041] The present application finds the suspected areas adjacent to the unqualified heat exchange area and adds the suspected areas to the calculation of the heat exchanger qualification rate, thereby issuing an early warning to the customer before the suspected areas are damaged.

[0042] like Figure 2 As shown, the present application also provides a heat exchanger fault diagnosis and predictive maintenance system based on data analysis, including a collection unit 100 and a data processing unit 200.

[0043] The working data of the heat exchanger is collected by the collection unit 100, and the data processing unit 200 executes the heat exchanger fault diagnosis and predictive maintenance method based on data analysis as described in any one of the embodiments of the first embodiment. The data processing unit 200 includes a data processing subunit 201 and a control subunit 202. The working data of the heat exchanger is processed by the data processing subunit 201, and the working state of the heat exchanger is controlled by the control subunit 202.

[0044] It should be noted that, for a heat exchanger, each heat exchange area corresponds to at least one collection unit 100. For example, for a heat exchanger including multiple heat exchange tubes, at least one collection unit 100 is provided for each heat exchange tube. The collection unit 100 includes multiple sensors, which can be used to collect working data of multiple different collection items in the heat exchange area. All collected working data are transmitted to the data processing unit 200. The data processing unit 200 includes a standard model of the heat exchanger, so as to determine whether an early warning needs to be issued based on the input working data of the heat exchanger.

[0045] The embodiments of the present invention are described in detail above with reference to the accompanying drawings, but the present invention is not limited thereto, and various changes can be made within the knowledge scope of technicians in the relevant technical field without departing from the purpose of the present invention.

Claims

1. A heat exchanger fault diagnosis and predictive maintenance method based on data analysis, characterized in that: include: Setting acquisition parameters, acquiring working data of multiple heat exchangers based on the acquisition parameters, preprocessing the working data of the heat exchangers to obtain preprocessed working data, and setting a qualified interval of the heat exchanger according to the preprocessed working data; Creating a heat exchanger standard model, and training the heat exchanger standard model by preprocessing the working data, so that the heat exchanger standard model determines the working state of the heat exchanger based on the working data of the heat exchanger, and obtaining the trained heat exchanger standard model; Collecting real-time data of the heat exchanger, inputting the real-time data of the heat exchanger into the trained standard model of the heat exchanger to obtain a real-time evaluation of the heat exchanger, and judging whether the qualified rate of the heat exchanger is greater than or equal to the qualified rate threshold based on the real-time evaluation of the heat exchanger; If the qualified rate of the heat exchanger is greater than or equal to the qualified rate threshold, the suspected area is obtained and the qualified rate trend of the heat exchanger is calculated by combining the suspected area with the unqualified area, and whether an early warning needs to be issued is determined based on the qualified rate trend of the heat exchanger; Select an unqualified heat exchange area; Obtaining the adjacent heat exchange regions of the unqualified heat exchange region; Determine whether the working status of adjacent heat exchange areas is qualified; If the working state of the adjacent heat exchange area is qualified, the adjacent heat exchange area is recorded as an observation area, and the historical working data of the observation area is obtained; Set the error threshold; Calculate the error between the historical working data of the observation area and the historical working data of the corresponding unqualified heat exchange area, and determine whether the error between the historical working data of the observation area and the historical working data of the corresponding unqualified heat exchange area is greater than or equal to an error threshold; If the error between the historical working data of the observed area and the historical working data of the corresponding unqualified heat exchange area is less than or equal to the error threshold, the observed area is recorded as a suspected area; Return and select an unqualified heat exchange area until all unqualified heat exchange areas are selected, and multiple suspected areas are obtained.

2. The method for heat exchanger fault diagnosis and predictive maintenance based on data analysis according to claim 1, characterized in that: Setting acquisition parameters, acquiring working data of multiple heat exchangers based on the acquisition parameters, preprocessing the working data of the heat exchangers to obtain preprocessed working data, and setting a qualified interval of the heat exchanger according to the preprocessed working data, including: Create a heat exchanger data table; Divide each heat exchanger into multiple heat exchange areas; The acquisition parameters are respectively set for the multiple heat exchange areas of each heat exchanger; the acquisition parameters include acquisition period, acquisition frequency and acquisition position; For each heat exchanger, the working data of multiple heat exchange areas of the heat exchanger are collected according to the collection parameters of the heat exchanger, and the collected working data are put into the heat exchanger data table; the working data includes multiple collection items of heat exchanger temperature data, heat exchanger pressure data and heat exchanger flow data.

3. The method for heat exchanger fault diagnosis and predictive maintenance based on data analysis according to claim 2, characterized in that: Setting acquisition parameters, acquiring working data of multiple heat exchangers based on the acquisition parameters, preprocessing the working data of the heat exchangers to obtain preprocessed working data, and setting a qualified interval of the heat exchanger according to the preprocessed working data, further comprising: Randomly select a heat exchanger from the heat exchanger data table and obtain the working data of the heat exchanger; Select a collection item and calculate the average value of the collection item; For this collection project, determine whether there is working data under each collection node in order according to the order of collection time; If there is no working data under the collection node, the average value of the collection item is filled into the collection node to complete the working data; Set a deviation threshold, calculate the upper and lower boundary values ​​based on the mean calculation of the collection item and the deviation threshold, and set the qualified interval based on the upper and lower boundary values; Return to select a collection item and calculate the average value of the collection item until the qualified interval of each collection item of the heat exchanger is obtained; Return to randomly select a heat exchanger from the heat exchanger data table until all heat exchangers in the working data table are selected, and obtain the qualified interval of each collection item of all heat exchangers.

4. The method for heat exchanger fault diagnosis and predictive maintenance based on data analysis according to claim 3 is characterized in that: Create a heat exchanger standard model, train the heat exchanger standard model by preprocessing the working data, so that the heat exchanger standard model can judge the working state of the heat exchanger based on the working data of the heat exchanger, and obtain the trained heat exchanger standard model, including: Create a standard model of a heat exchanger; Divide all the pre-processed working data of the heat exchanger into training set and test set in random proportion; The training set is input into the heat exchanger standard model, so that the heat exchanger standard model continuously learns the coupling relationship between the heat exchanger working data and the heat exchanger state, and obtains the trained heat exchanger standard model; The test set is input into the trained heat exchanger standard model to verify whether the trained heat exchanger standard model is trained.

5. The method for heat exchanger fault diagnosis and predictive maintenance based on data analysis according to claim 4, characterized in that: The training set is input into the heat exchanger standard model, so that the heat exchanger standard model continuously learns the coupling relationship between the heat exchanger working data and the heat exchanger state, and the trained heat exchanger standard model is obtained, including: For each heat exchanger in the training set, obtain the working data of each heat exchanger respectively; Select a heat exchanger and determine in turn whether the working data of each collected item of the heat exchanger is within the qualified range; If the working data of the collected items is within the qualified range, the heat exchanger is recorded as qualified; If the working data of the collected items is not within the qualified range, the heat exchanger is recorded as unqualified; Return to select a heat exchanger until all heat exchangers are selected and the working status of each heat exchanger is obtained; Based on the working data of each heat exchanger and the working state under the working data, a coupling relationship of heat exchange area-working data-working state is established, and the heat exchanger, the working data, and the coupling relationship of heat exchange area-working data-working state are used as a training sample, thereby obtaining multiple training samples; Multiple training samples are sequentially input into the heat exchanger standard model, so that the heat exchanger standard model continuously learns the corresponding relationship between the heat exchanger working data and the heat exchanger state, and obtains the trained heat exchanger standard model.

6. The method for heat exchanger fault diagnosis and predictive maintenance based on data analysis according to claim 5, characterized in that: Collect the real-time data of the heat exchanger, input the real-time data of the heat exchanger into the trained standard model of the heat exchanger, obtain the real-time evaluation of the heat exchanger, and judge whether the qualified rate of the heat exchanger is greater than or equal to the qualified rate threshold based on the real-time evaluation of the heat exchanger, including: Collect real-time data of each heat exchange area of ​​the heat exchanger; Input the real-time data of each heat exchange area of ​​the heat exchanger into the trained standard model of the heat exchanger, and obtain the real-time evaluation of each heat exchange area of ​​the heat exchanger output by the trained standard model of the heat exchanger; Set pass rate threshold; Determine in turn whether the real-time evaluation of each heat exchange area of ​​the heat exchanger is qualified, and calculate the qualified rate of all heat exchange areas of the heat exchanger; Determine whether the qualified rate of all heat exchange areas of the heat exchanger is greater than or equal to the qualified rate threshold; If the qualified rate of a heat exchange area of ​​the heat exchanger is lower than the qualified rate threshold, the heat exchanger is stopped and an early warning is issued.

7. The method for heat exchanger fault diagnosis and predictive maintenance based on data analysis according to claim 6, characterized in that: If the qualified rate of the heat exchanger is greater than or equal to the qualified rate threshold, the suspected area is obtained and the qualified rate trend of the heat exchanger is calculated by combining the suspected area with the unqualified area. Based on the qualified rate trend of the heat exchanger, it is determined whether an early warning needs to be issued, including: If the qualified rate of all heat exchange areas of the heat exchanger is greater than or equal to the qualified rate threshold, it is determined whether there is an unqualified heat exchange area; If there is no unqualified heat exchange area, keep the heat exchanger working; If there is an unqualified heat exchange area, a suspected area is obtained according to the unqualified heat exchange area.

8. The method for heat exchanger fault diagnosis and predictive maintenance based on data analysis according to claim 7, characterized in that: If the qualified rate of the heat exchanger is greater than or equal to the qualified rate threshold, the suspected area is obtained and the qualified rate trend of the heat exchanger is calculated by combining the suspected area with the unqualified area. Based on the qualified rate trend of the heat exchanger, it is determined whether an early warning needs to be issued, which also includes: The qualified rate of the heat exchanger is recalculated by combining the suspected area with the unqualified heat exchange area to obtain the trend qualified rate of the heat exchanger; Determine whether the trend qualification rate of the heat exchanger is greater than or equal to the qualification rate threshold; If the trend qualified rate of the heat exchanger is greater than or equal to the qualified rate threshold, the operation of the heat exchanger is maintained; If the trend pass rate of the heat exchanger is less than the pass rate threshold, an early warning is issued.

9. A heat exchanger fault diagnosis and predictive maintenance system based on data analysis, characterized in that: include: A collection unit, through which the working data of the heat exchanger is collected; A data processing unit, through which the method for heat exchanger fault diagnosis and predictive maintenance based on data analysis as described in any one of claims 1 to 8 is executed, the data processing unit includes a data processing subunit and a control subunit, the working data of the heat exchanger is processed by the data processing subunit, and the working state of the heat exchanger is controlled by the control subunit.

Images