CN115289606A - Dehumidifier online fault diagnosis method, system, server and storage medium - Google Patents
Dehumidifier online fault diagnosis method, system, server and storage medium Download PDFInfo
- Publication number
- CN115289606A CN115289606A CN202210820907.7A CN202210820907A CN115289606A CN 115289606 A CN115289606 A CN 115289606A CN 202210820907 A CN202210820907 A CN 202210820907A CN 115289606 A CN115289606 A CN 115289606A
- Authority
- CN
- China
- Prior art keywords
- data
- point
- fault
- target point
- dehumidifier
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000003745 diagnosis Methods 0.000 title claims abstract description 36
- 238000000034 method Methods 0.000 title claims abstract description 35
- 238000003860 storage Methods 0.000 title claims abstract description 10
- 238000010801 machine learning Methods 0.000 claims abstract description 64
- 238000012545 processing Methods 0.000 claims abstract description 31
- 238000000605 extraction Methods 0.000 claims abstract description 26
- 230000002159 abnormal effect Effects 0.000 claims abstract description 13
- 238000004364 calculation method Methods 0.000 claims description 28
- 238000004422 calculation algorithm Methods 0.000 claims description 25
- 238000012549 training Methods 0.000 claims description 20
- 238000012360 testing method Methods 0.000 claims description 18
- 238000004590 computer program Methods 0.000 claims description 11
- 238000001310 location test Methods 0.000 claims description 6
- 239000000126 substance Substances 0.000 claims description 4
- 238000007791 dehumidification Methods 0.000 abstract description 34
- 238000001514 detection method Methods 0.000 abstract description 12
- 238000005516 engineering process Methods 0.000 abstract description 7
- 238000004519 manufacturing process Methods 0.000 abstract description 3
- 238000004458 analytical method Methods 0.000 abstract description 2
- 230000006870 function Effects 0.000 description 19
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 12
- 238000010586 diagram Methods 0.000 description 10
- 230000008569 process Effects 0.000 description 7
- 230000005856 abnormality Effects 0.000 description 5
- 230000008859 change Effects 0.000 description 5
- 230000008929 regeneration Effects 0.000 description 5
- 238000011069 regeneration method Methods 0.000 description 5
- 230000005540 biological transmission Effects 0.000 description 4
- 238000007405 data analysis Methods 0.000 description 4
- 238000012423 maintenance Methods 0.000 description 4
- 238000012795 verification Methods 0.000 description 4
- 238000007726 management method Methods 0.000 description 3
- 238000010276 construction Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000007613 environmental effect Effects 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 238000007689 inspection Methods 0.000 description 2
- 230000003993 interaction Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000009423 ventilation Methods 0.000 description 2
- 230000009471 action Effects 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 238000004140 cleaning Methods 0.000 description 1
- 238000011217 control strategy Methods 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 238000013523 data management Methods 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 238000003066 decision tree Methods 0.000 description 1
- 230000004069 differentiation Effects 0.000 description 1
- 238000005265 energy consumption Methods 0.000 description 1
- 230000008014 freezing Effects 0.000 description 1
- 238000007710 freezing Methods 0.000 description 1
- 238000010438 heat treatment Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000007637 random forest analysis Methods 0.000 description 1
- 230000001172 regenerating effect Effects 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
Images
Classifications
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F24—HEATING; RANGES; VENTILATING
- F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
- F24F11/00—Control or safety arrangements
- F24F11/30—Control or safety arrangements for purposes related to the operation of the system, e.g. for safety or monitoring
- F24F11/32—Responding to malfunctions or emergencies
- F24F11/38—Failure diagnosis
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F24—HEATING; RANGES; VENTILATING
- F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
- F24F11/00—Control or safety arrangements
- F24F11/50—Control or safety arrangements characterised by user interfaces or communication
- F24F11/61—Control or safety arrangements characterised by user interfaces or communication using timers
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F24—HEATING; RANGES; VENTILATING
- F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
- F24F11/00—Control or safety arrangements
- F24F11/62—Control or safety arrangements characterised by the type of control or by internal processing, e.g. using fuzzy logic, adaptive control or estimation of values
- F24F11/63—Electronic processing
- F24F11/64—Electronic processing using pre-stored data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
Landscapes
- Engineering & Computer Science (AREA)
- General Engineering & Computer Science (AREA)
- Mechanical Engineering (AREA)
- Combustion & Propulsion (AREA)
- Chemical & Material Sciences (AREA)
- Mathematical Physics (AREA)
- Signal Processing (AREA)
- Software Systems (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Medical Informatics (AREA)
- General Physics & Mathematics (AREA)
- Computing Systems (AREA)
- Human Computer Interaction (AREA)
- Evolutionary Computation (AREA)
- Data Mining & Analysis (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Fuzzy Systems (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Testing And Monitoring For Control Systems (AREA)
Abstract
The application provides a dehumidifier online fault diagnosis method, a dehumidifier online fault diagnosis system, a dehumidifier online fault diagnosis server and a dehumidifier online fault diagnosis storage medium, wherein the method comprises the following steps: acquiring point location data representing the working state of the point location and a fault target point to be diagnosed, acquiring other point locations influencing the fault target point according to the fault target point, and acquiring other point location data representing the working states of other point locations based on the other point locations and the point data; performing feature extraction operation on other point location data to obtain a feature value; and processing the characteristic value by using a machine learning model to obtain the prediction data of the fault target point, and judging the running state of the dehumidifier based on the prediction data and the real-time data of the fault target point. Through the method and the device, operations such as analysis, modeling and operation are performed on the off-line data and the on-line data of the dehumidification system, the problems that the dehumidification system is not timely and inaccurate in fault detection and consumes manpower through the traditional technology are solved, and the production or storage loss caused by abnormal temperature and humidity of the space controlled by the dehumidification system is reduced.
Description
Technical Field
The application relates to the technical field of online fault diagnosis, in particular to a dehumidifier online fault diagnosis method, a dehumidifier online fault diagnosis system, a dehumidifier online fault diagnosis server and a dehumidifier online fault diagnosis storage medium.
Background
The dehumidification system is characterized in that an indoor air return port pumps indoor air with moisture into a main machine, the indoor air return port pumps the moisture in the air into the main machine through a compressor, and then the air is discharged indoors through an indoor air supply port, indoor air circulation is completed through two air ports of the indoor air return port and the indoor air supply port, the humidity of indoor air is adjusted, and people are enabled to be in a comfortable space. However, because of the large scale and the complex control strategy, various faults occur frequently.
Common abnormal faults of the dehumidification system include valve faults, pipeline faults, heat exchange coil faults, dehumidification rotating wheel faults, motor faults, fan faults and control system faults. In recent years, researchers have proposed a plurality of dehumidification system fault detection technologies, mainly including four fault detection technologies, namely, manual field inspection; the dehumidification system has an abnormal alarm function; when the field data is accessed to the data center, the early warning function provided by an online detection platform in a control room or the data center; and a worker observes the change trend of the online data in a control room or a data center, observes early warning provided by an online detection platform and finally makes fault judgment. However, these existing dehumidification system fault detection techniques have some disadvantages, for example, when the number of dehumidification systems in a plant area is as many as hundreds, a lot of manpower is needed to perform inspection; not all the devices in the system have the self-alarming function, or part of system faults and abnormalities are missed; the early warning mode is simple, only the mode of upper and lower threshold values is adopted, the result of the single-measuring-point data or the multi-measuring-point data which is calculated according to the rule is judged, the alarm is given when the result exceeds the set threshold value, and the threshold value cannot be automatically modified after being set; the process of analyzing observable alarms, data change trends and the like by workers needs a certain time, so that faults can not be judged in time. Namely, the fault and abnormality detection of the dehumidification system cannot be timely, accurately and effectively finished, so that the space temperature and humidity abnormal state controlled by the dehumidification system causes loss to production and storage.
Disclosure of Invention
The problems that fault detection is not timely and inaccurate and manpower is consumed by the dehumidification system through the traditional technology are solved by analyzing, modeling, calculating and the like on the off-line data and the on-line data of the dehumidification system, and the production or storage loss caused by abnormal temperature and humidity of the space controlled by the dehumidification system is reduced.
In a first aspect, the present embodiment provides an online fault diagnosis method for a dehumidifier, where the method includes:
acquiring point location data representing the working state of the point location and a fault target point to be diagnosed, acquiring other point locations influencing the fault target point according to the fault target point, and acquiring other point location data representing the working state of other point locations based on the other point locations and the point location data;
performing feature extraction operation on the other point location data to obtain a feature value; and processing the characteristic value by using a machine learning model to obtain the prediction data of the fault target point, and judging the running state of the dehumidifier based on the prediction data and the actual data of the fault target point.
In some embodiments, the obtaining of other points affecting the fault target point according to the fault target point includes:
and judging the fault type of the fault target point according to the fault target point, and determining other point locations influencing the fault target point based on the fault type, wherein each fault type has a judgment standard for determining other point locations influencing the fault target point.
In some of these embodiments, the feature extraction operation comprises at least one of: and performing operation on one point location data in the other point location data within a period of time, and performing operation on a plurality of point location data in the other point location data within a period of time.
In some embodiments, the processing the feature values using a machine learning model to obtain the prediction data of the fault target point includes:
acquiring equipment parameters, fault target points and characteristic values of the dehumidifier at regular time, and acquiring historical actual data of the fault target points and historical characteristic values of other points influencing the fault target points according to the fault target points;
determining an initial machine learning model for processing the characteristic value based on the equipment parameters of the dehumidifier and the fault target point, training the initial machine learning model in an off-line manner according to the historical characteristic value and the historical actual data of the fault target point to obtain a final machine learning model, and processing the other point data in an on-line manner by using the final machine learning model to obtain the prediction data of the fault target point.
In some embodiments, the training the initial machine learning model offline according to the historical feature values and the historical actual data of the fault target point, and obtaining the final machine learning model further includes:
acquiring other point location test data and fault target point test data which are used for testing the accuracy of the final machine learning model in the historical characteristic values and the historical actual data of the fault target point, and processing the other point location test data by using the final machine learning model to acquire test prediction data of the fault target point;
judging whether the difference value between the test prediction data of the fault target point and the actual data of the fault target point is larger than a threshold value, if so, adjusting the feature extraction operation and/or selecting a new initial machine learning model; otherwise, a final machine learning model is obtained.
In some embodiments, the determining the operating state of the dehumidifier based on the predicted data and actual data of a fault target point includes:
acquiring a deviation between the prediction data and actual data of a fault target point, recording the number of times that the deviation exceeds a preset deviation within fixed time, judging whether the number of times exceeds a preset number of times, if so, judging that the running state of the dehumidifier is abnormal, and sending an alarm signal by the dehumidifier; otherwise, the running state of the dehumidifier is normal.
In a second aspect, the present embodiment provides an online fault diagnosis system for a dehumidifier, where the system includes a data processing module and an algorithm calculation module; wherein the content of the first and second substances,
the data processing module is used for acquiring point location data representing the working state of the point location;
the algorithm calculation module is used for acquiring a fault target point to be diagnosed, acquiring other point locations influencing the fault target point according to the fault target point, and acquiring other point location data representing the working states of other point locations based on the other point locations and the point location data; performing feature extraction operation on the other point location data to obtain a feature value; and processing the characteristic value by using a machine learning model to obtain the prediction data of the fault target point, and judging the running state of the dehumidifier based on the prediction data and the actual data of the fault target point.
In some of these embodiments, the system further comprises a software front-end module and a software back-end module; wherein the content of the first and second substances,
the software front-end module is used for displaying the running state of the dehumidifier and the prediction data of the fault target point;
the software back-end module is used for interacting with the data processing module, the algorithm calculation module and the software front-end module.
In a third aspect, an embodiment of the present application provides a server, where the server includes: a processor and a memory, wherein the memory stores a computer program capable of running on the processor, and when the computer program is executed by the processor, the online fault diagnosis method for the dehumidifier is realized.
In a fourth aspect, the present application provides a computer-readable storage medium, on which a computer program capable of running on a processor is stored, wherein the computer program, when executed by the processor, implements the dehumidifier online fault diagnosis method according to the first aspect.
By adopting the scheme, when the fault of the dehumidifier is diagnosed on line, the fault type of the fault target point is judged according to the fault target point, other point location data which directly influence the data change of the fault target point are determined based on the fault type, the operation for feature extraction is selected in a targeted manner according to the fault type, the feature value is obtained, and the feature value can more accurately represent the data features of other point locations by performing combined operation on multiple point locations; then, determining an initial machine learning model for processing characteristic values based on equipment parameters of the dehumidifier and fault target points, training the initial machine learning model in an off-line mode according to historical data of other point locations and historical actual data of the fault target points to obtain a final machine learning model, and processing the data of the other point locations on line by using the final machine learning model to obtain predicted data of the fault target points; and finally, obtaining a difference value of the prediction data and the actual data according to the prediction data and the actual data, recording the times that the difference value exceeds the preset deviation in fixed time, if the times do not exceed the preset times, determining that the dehumidifier normally operates, and if not, determining that the dehumidifier has a fault.
According to the method and the system, on one hand, manpower required by fault diagnosis of the dehumidification system is saved, on the other hand, abnormal alarm of equipment, fault diagnosis of the equipment and dynamic early warning of equipment data trend are realized through data analysis and machine learning technologies, and the comprehensive application of the multiple algorithm functions can also improve the accuracy of fault alarm. In addition, comprehensive data analysis, modeling and operation are carried out on the fault target point and the point positions related to the fault target point, so that possible abnormalities are comprehensively and comprehensively analyzed, and the fault identification rate is improved. Therefore, the scheme can timely, accurately and effectively complete the fault and abnormity detection of the dehumidification system.
Drawings
Fig. 1 is a schematic diagram of an operation of a dual-rotor dehumidification system provided in this embodiment.
Fig. 2 is a block diagram of a dehumidifier fault diagnosis system provided in this embodiment.
Fig. 3 is a flowchart of an online fault diagnosis method for a dehumidifier according to this embodiment.
Fig. 4 is a flowchart of a dehumidifier early warning scheme based on machine learning according to this embodiment.
Fig. 5 is a schematic diagram of maintenance of a machine learning model according to this embodiment.
Fig. 6 is an explanatory diagram of a part of dots in the dehumidification system provided in this embodiment.
FIG. 7 is a comparison of the predicted results of the training set for the test provided in this example.
Fig. 8 is an enlarged view of the prediction stage of fig. 7 provided in the present embodiment.
Fig. 9 is a block diagram of the server according to the present embodiment.
Detailed Description
For a clearer understanding of the objects, technical solutions and advantages of the present application, reference is made to the following description and accompanying drawings. However, it will be apparent to one of ordinary skill in the art that the present application may be practiced without these specific details. It will be apparent to those of ordinary skill in the art that various changes can be made to the embodiments disclosed herein, and that the general principles defined herein may be applied to other embodiments and applications without departing from the principles and scope of the present application. Thus, the present application is not limited to the embodiments shown, but is to be accorded the widest scope consistent with the scope of the present application as claimed.
The embodiments of the present application will be described in further detail with reference to the drawings.
The double-rotary-wheel dehumidification system comprises a plurality of subsystems: the system comprises a regeneration system, a ventilation system, a steam system, a frozen water system, a drainage system, an electric power system and a control system, wherein each subsystem consists of certain equipment or components. The regeneration system comprises two dehumidification rotating wheels, namely a primary rotating wheel and a secondary rotating wheel, a fan, a motor, an air pipe, a heat exchange coil and the like; the ventilation system comprises a fan, a motor, an air pipe, a valve and the like; the power system comprises a power line, a circuit breaker, a power cabinet and the like; the control system comprises a PLC controller, a frequency converter, a sensor, a data acquisition and man-machine interaction interface and the like.
Fig. 1 is a schematic diagram of an operation of a dual-rotor dehumidification system provided in this embodiment. As shown in fig. 1, when the inlet air and the return air pass through the frozen water coil, the frozen water calandria can change the temperature of the air passing through the position of the frozen water coil, so that the heat exchange between the air and the frozen water is realized, the heat in the air is reduced, the temperature of the air is reduced, the air in a certain temperature range is finally formed to pass through the rotating wheel, and the temperature of the air passing through the rotating wheel is ensured to meet the process requirement; when the air through the runner is normal temperature, the runner has the ability of moisture in the adsorbed air, therefore the inlet air and return air are when freezing water coil pipe and one-level and/or second grade runner, and air humidity is lower, and the moisture that carries in the air is adsorbed by the runner, and air humidity after the runner descends, and the dry air who obtains sends into the space of double-runner dehumidification system control to make the humidity of being controlled the space keep in the within range of settlement. Meanwhile, part of the dry air formed by the air entering after passing through the rotating wheels twice enters a regeneration loop, so that the temperature of the regenerated air is improved through the steam coil, when the air passing through the rotating wheels is high temperature, the rotating wheels can release moisture, the rotating wheels are locally dried, and finally the air containing the moisture is discharged through the air outlet.
In addition, the rotating wheel is driven by the motor to rotate, and when passing through an air inlet loop and an air return loop, any position on the rotating wheel takes away moisture in the air, and then the moisture reaches a regeneration loop under the action of rotation, is dried at the regeneration loop, and releases the adsorbed moisture. Through continuous rotation, the system achieves dynamic balance, partial moisture in the air entering the air and the air returning is brought out of the system by the air exhausted from the air, and finally, the moisture in the air supplied is less than that of the air entering the air.
The primary rotating wheel and the secondary rotating wheel are used for changing the air humidity; the frozen water coil and the steam coil are used for changing the temperature of air passing through the positions of the frozen water coil and the steam coil, so that the temperature of the air passing through the rotating wheel can meet the process requirements, and the flow of the frozen water or the steam passing through the frozen water coil and the steam coil can be controlled by respective valves, so that the effects of flexibly adjusting the temperature of the frozen water and heating the steam can be achieved; the fan is used for controlling the direction and the speed of the airflow and is driven by the motor to operate.
In the embodiment, the return air comes from the space controlled by the double-rotor dehumidification system, so that on one hand, the temperature and humidity conditions of the controlled space can be measured at the return air position, and the control system is convenient to adjust; on the other hand, the air circulation in the controlled space can be ensured, and the temperature and the humidity of return air are relatively close to the control target under the normal condition, so that the energy consumption of the system is reduced.
The embodiment provides an online fault diagnosis system for a dehumidifier, and fig. 2 is a structural block diagram of the online fault diagnosis system for the dehumidifier provided by the embodiment. As shown in fig. 2, the system includes: the system comprises a data processing module, a software back-end module, an algorithm calculation module and a software front-end module.
The data processing module is used for acquiring point location data representing the working state of the point location.
The data processing module comprises three types of devices including sensors, data acquisition and data transmission, the number and the topological structure of the devices can be determined according to the actual physical location condition, and the embodiment is not limited. The point location data required by the system corresponds to output data of the sensor, for example, a temperature sensor is arranged at the air supply outlet to correspond to the point location data of the temperature of the air supply outlet. One sensor device can measure one or more data at a point location, such as a temperature and humidity integrated sensor, and can simultaneously measure the temperature and humidity at a point location to generate two data. The sensors need to measure key operation data, process data, replacement data and the like of the dehumidifier in real time when in work. Such as temperature and humidity at various critical locations in the dehumidification system piping; the temperature and humidity within the controlled space; opening degree of each valve in the dehumidification system; the rotating speed of the dehumidifying rotating wheel; the temperature and vibration of each bearing of the fan and the motor in the dehumidification system and the like are detected by corresponding sensors.
After the point location is measured by the sensor, the point location needs to be collected by data acquisition equipment, the continuous detection result of the sensor is converted into discrete data with a certain period, and the period can be adjusted according to the algorithm requirement or other conditions. Usually, one sensor device is provided with one data acquisition device, or a plurality of sensor devices may share one data acquisition device. Under the condition that a plurality of sensor devices share one data acquisition device, the sensors which are relatively close to each other can be selected to share the data acquisition device, and the sensor devices which share the data acquisition device can also be selected according to the actual condition.
After the data is collected by the data collection device, the data is transmitted by the data transmission device, such as a router, a gateway and the like. Generally, data acquisition devices in the same physical space can share one set of data transmission devices, and the specific implementation is determined by the conditions of the actual physical space.
The software back-end module is used for interacting data with other modules, managing data obtained by each module, managing task progress or thread of each part and the like, and is a center for connecting all the modules. The method specifically comprises but is not limited to a data acquisition function, a data storage and management function, a simple data processing function, a core algorithm task calling and management function and a front-end interaction function.
The method and the device can be used for storing data in the data processing module and calculation results of the algorithm calculation module and sending the data and the calculation results to the software front-end module.
The algorithm calculation module is used for acquiring a fault target point to be diagnosed, acquiring other point locations influencing the fault target point according to the fault target point, and acquiring other point location data representing the working states of the other point locations based on the other point locations and the point data; performing feature extraction operation on other point location data to obtain a feature value; and processing the characteristic value by using a machine learning model to obtain the prediction data of the fault target point, and judging the running state of the dehumidifier based on the prediction data and the actual data of the fault target point.
According to the potential fault occurrence probability, selecting a fault type for further diagnosis from potential fault types existing in the dehumidification system, and determining a fault target point based on the fault type, for example, selecting the fault type for further diagnosis as the temperature of the air supply outlet, and then the fault target point is the air supply outlet. And the accurate fault diagnosis of the dehumidification system is comprehensively realized through a machine learning technology by combining the operation data of other point positions associated with the fault target point and combining the operation mechanism and the environmental condition of the equipment.
The software front-end module is mainly used for interacting with a user using software, displaying data in the system in a certain form and simultaneously obtaining operation information of the user. Specific software front-end functions include, but are not limited to: the system comprises an operation interface function, a user login and management function, a system overview function, an equipment real-time data display function, a fault alarm function, an alarm display and management function, a historical data display function, a system setting function and the like. The method and the device can be particularly used for displaying the running state of the dehumidifier and the prediction data of the fault target point.
In the dehumidification system of the embodiment, the data processing module performs data detection, acquisition and transmission of a plurality of point locations on the dehumidification system; the rear end part of the software receives the data, stores the data, and calls an algorithm calculation module to calculate the data after processing the data to a certain extent; the algorithm calculation module returns the calculation result to the software back-end module; the software rear end part sends the obtained data and the calculation result to the software front end module, and the software front end module is responsible for displaying; the hardware platform included in the embodiment is a server carrying a software back-end module and an algorithm calculation module; the software front-end module can be displayed on a local hardware platform and also can be displayed on other hardware platforms through a network, such as remote display on a computer; the user can set and control the fault diagnosis system through the software front-end module or the remote software front-end module, the operation information of the user in the software front-end module is sent to the software rear-end module, and the software rear-end module correspondingly adjusts the parameters or the structure of the core algorithm or the software rear-end module according to the information input by the user.
Based on the system, the system provides an online fault diagnosis method for the dehumidifier. Fig. 3 is a flowchart of the online fault diagnosis method for a dehumidifier provided in this embodiment. As shown in fig. 3, the process includes the following steps:
step S301, point location data representing the working state of the point location and a fault target point needing to be diagnosed are obtained, other point locations influencing the fault target point are obtained according to the fault target point, and other point location data representing the working state of other point locations are obtained based on other point locations and point data.
In this embodiment, the sensor records the point location data of different positions of the dehumidifier in real time, and can obtain other point location data representing the working states of other point locations from the point location data according to other point locations,
obtaining other point locations influencing the fault target point according to the fault target point comprises the following steps: and judging the fault type of the fault target point according to the fault target point, and determining other point positions influencing the fault target point based on the fault type, wherein each fault type has a judgment standard for determining other point positions influencing the fault target point.
The fault types of the fault target points in the embodiment include a fault type influenced by the wind direction of the pipeline and a fault type of the operating parameters of the equipment in the dehumidification system. The factors affecting each type of fault are different, the factor affecting the type of fault affected by the wind direction in the duct being the wind direction in the duct, and the factor affecting the type of fault in the operating parameters of the device being other data of the device. For example, the fault target point belongs to a fault type influenced by the wind direction of the pipeline, and the specific operation of determining other points influencing the fault target point based on the fault type is as follows: and determining a first pipeline where the fault target point is located according to the fault target point, and determining first other points in the first pipeline, which directly influence the fault target point, based on the gas flow direction of the first pipeline, wherein the first other points at least comprise one point. In this embodiment, two pipelines are provided, so that it is further necessary to determine whether the gas flows into the second pipeline from the first other point location to the fault target point location when encountering the bifurcation, and if so, determine a second other point location directly affecting the fault target point in the second pipeline according to the flow direction of the gas, where the other point locations include the first point location and the second point location; if there is no flow into the second pipe, the second other point is empty, and the other point is the first other point. And when the fault target point is diagnosed, the influence of other point locations on the data at the fault target point is also fully considered, and linkage analysis is carried out on the data and other point locations.
Step S302, performing feature extraction operation on other point location data to obtain a feature value; and processing the characteristic value by using a machine learning model to obtain the prediction data of the fault target point, and judging the running state of the dehumidifier based on the prediction data and the actual data of the fault target point.
When feature extraction is performed on other point location data, at least one of the following operation modes can be adopted: operating certain point data in other point data within a period of time; and operating a plurality of point data in other point data in a period of time. The operation of a certain point data in other point data in a period of time may be components of energy after mean, maximum, minimum, variance, standard deviation, integration, differentiation and fourier transform of the point data in the period of time, and the operation of several point data in other point data in a period of time may be operations of addition, product, difference, proportion, mean, variance, standard deviation and the like of data of the point in a period of time.
The characteristic value after the operation is carried out on the point data can more simply and clearly show the change trend of the point data, and the characteristic of the current point is reflected. Therefore, the characteristic value obtained after the operation is carried out on the point data is used for replacing the original point data according to the actual situation, and the fault target point and the point related to the fault target point are subjected to comprehensive data analysis, so that possible abnormalities are comprehensively and comprehensively analyzed, and the fault identification rate is improved.
Fig. 4 is a flowchart of a dehumidifier early warning scheme based on machine learning according to this embodiment. As shown in fig. 4, the scheme includes target establishment, data acquisition, feature extraction, model construction, online operation, and model maintenance. Wherein the content of the first and second substances,
the target establishment is used for determining the fault type to be predicted or classified, and determining a fault target point. The data acquisition is used for acquiring operation data, environmental data and process data, performing pretreatment such as data cleaning on the data and discarding the data with errors in detection. Feature extraction is used for empirically combing predicted fault target points, such as which faults are easy to occur to the equipment and which corresponding fault causes are, and meanwhile, other data points are subjected to statistical analysis to find key features related to the faults. The steps S301 and S302 have already been described in detail, and are not described again here.
According to the characteristics, types, data conditions, characteristic conditions and other factors of the fault target points, a proper machine learning algorithm needs to be selected for model construction. After the algorithm is selected, the model needs to be trained, and meanwhile, through adjustment of algorithm parameters, an ideal calculation result is achieved. And finally, evaluating the calculation result of the model, and if the calculation result does not meet the requirements, readjusting the parameters of the model, or adjusting the characteristics, or reselecting the model, so that the step needs to interact with the characteristic extraction, and the final ideal calculation result is achieved through repeated characteristic extraction and model training.
Since the dehumidifier has respective equipment parameters and the fault target point has respective characteristics, the initial machine learning model needs to be determined based on the equipment parameters and the fault target point of the dehumidifier, and then the initial machine learning model is trained offline according to the historical characteristic value and the historical actual data of the fault target point to obtain the final machine learning model.
In addition, the model is constructed and interacted with the feature extraction, and a final ideal calculation result is achieved through repeated feature extraction and model training, namely, the initial machine learning model is trained in an off-line mode according to historical feature values and historical actual data of fault target points, and a final machine learning model is obtained. The machine learning model includes many specific algorithms, and the specific machine learning algorithms include but are not limited to: BP neural networks, decision trees, support vector machines, K neighbors, random forests, LSTM, ensemble learning, XGboost, lightGBM, convolutional neural networks, and the like.
Before obtaining the final machine learning model, the method further comprises: acquiring other point location test data and fault target point test data used for testing the accuracy of the final machine learning model in the historical characteristic values and the historical actual data of the fault target point, and processing the other point location test data by using the final machine learning model to acquire test prediction data of the fault target point; judging whether the difference value between the test prediction data of the fault target point and the actual data of the fault target point is larger than a threshold value, if so, indicating that the current machine learning model is unavailable, adjusting the feature extraction operation and/or selecting a new initial machine learning model, then training, and stopping training until the machine learning model is available; otherwise, the current machine learning model is available, and the final machine learning model is obtained.
After the final machine learning model is obtained, the feature extraction operation mode includes the operation of a certain point data within a period of time, so that the feature extraction operation needs to be adjusted to a certain extent, and the off-line data feature extraction calculation is converted into the on-line data feature extraction calculation. The model after offline to online operation can be operated online, and the future data can be calculated in real time to obtain the predicted data of the fault target point.
After the model is operated online, a real-time result can be obtained, but at the same time, the performance of the model result also needs to be concerned, if the accuracy of the prediction result is compared with the actual situation, the time consumption of prediction calculation and the like, and new fault case data are taken as training samples periodically, and the original model is subjected to incremental learning, so that the accuracy of the model is further improved, or the situation that the performance of the model is reduced along with the time is prevented.
Fig. 5 is a schematic diagram of maintenance of a machine learning model according to this embodiment. As shown in fig. 5, the data obtained by real-time measurement is called online data, the online data is subjected to online feature extraction to obtain a plurality of quantized feature values, the feature values are used as the input of a machine learning model, and corresponding output, namely, the alarm or diagnosis result of an abnormality or a fault is obtained through the calculation of the model.
After the online data occurs, the online data is stored in a certain mode and becomes offline data. When the machine learning model is initially built or is updated subsequently, a sufficient amount of offline data needs to be utilized for model training. And (4) extracting the features of the offline data to obtain a plurality of historical feature values, wherein the historical feature values are used for training the model. After the model training is completed and the ideal effect is achieved on the verification data, the model can be used for on-line calculation. It should be noted that, when different fault diagnosis problems are solved, the process of the dehumidifier early warning scheme based on machine learning and the maintenance mode of the machine learning model are the same, but the point location data, the characteristics and the machine learning model which are specifically used are not necessarily the same, and need to be selected according to actual conditions.
In this embodiment, the determining the operating state of the dehumidifier based on the predicted data and the actual data of the fault target point includes: acquiring the deviation between the prediction data and the actual data of the fault target point based on the prediction data, recording the times that the deviation exceeds the preset deviation within fixed time, judging whether the times exceeds the preset times, if so, judging that the running state of the dehumidifier is abnormal, and sending an alarm signal by the dehumidifier; otherwise, the running state of the dehumidifier is normal. The device abnormity alarming, the device fault diagnosis and the device data trend dynamic early warning are realized through data analysis and machine learning technologies, and the comprehensive application of the multiple algorithm functions can also improve the fault alarming accuracy.
In this embodiment, after the operation state of the dehumidifier is abnormal, setting and/or controlling a fault target point to protect the dehumidifier is further included.
This embodiment also provides for a failure in a dual-rotor dehumidification system: and testing the air supply dew point abnormity by taking a machine learning model of a LightGBM algorithm as an example. Fig. 6 is an explanatory diagram of a part of dots in the dehumidification system provided in the present embodiment. As shown in fig. 6, each number represents point location data, (1) air supply dew point, (2) opening degree of a rear frozen water valve, (3) temperature of a middle air duct, (4) humidity of the middle air duct, (5) rotation speed of a middle fan, (6) rotation speed of a secondary runner, (7) opening degree of a rear steam valve, (8) temperature of a rear regenerative air duct, and (9) temperature of a rear exhaust air duct.
In the test, (1) is taken as a fault target point, and (2) to (9) are taken as other point position data required by the model. And then, performing feature extraction on the (2) to (9) in an average value mode, and adopting a machine learning model of a LightGBM algorithm, wherein the test totally adopts 15000 moments of data in a time period from 2021.10.07 to 2021.10.17, wherein the first 12500 moments are used as training and verification data, and the ratio of the training data to the verification data is 7:3, an algorithm randomly selects 70% of data from 12500 moments as training data and 30% of data as verification data. The remaining 2500 moments of data are used as a prediction set, which contains about 300 moments of abnormal data and about 2200 moments of normal data. Each time data comprises a plurality of characteristic values of point data from (2) to (9). FIG. 7 is a comparison of the prediction results of the training set provided in the test of the present embodiment. As shown in fig. 7, the gray dotted line portion is the real historical data used for model training, the gray solid line portion is the real historical data that has not been trained, and the black solid line portion is the result of prediction of the air supply dew point after the model has been trained. In the training stage, the prediction data can be accurately superposed with the real data, and the prediction error is
And the accuracy of the model is higher within 5 ℃.
Fig. 8 is an enlarged view of the prediction stage of fig. 7 provided in the present embodiment. As shown in fig. 8, in the prediction part of the future data, when the equipment is normal, the prediction data and the real data can be well overlapped, and when the equipment is abnormal or has a fault, the real data and the prediction data have a large deviation, the deviation of the air supply dew point prediction in the test reaches 30 ℃ at most, and exceeds the normal error range, the system gives an alarm, and the air supply dew point is considered to be abnormal. And finally, deploying the trained and verified model to an online fault diagnosis system of the dehumidifier, calling the model by the back end of the software, performing online and real-time calculation on future data, and sending a calculation result to the front end of the software for display.
Fig. 9 is a block diagram of a server provided in this embodiment, and as shown in fig. 9, the server includes a processor 91 and a memory 92, where the memory 92 stores a computer program 93 capable of running on the processor 91, and when the computer program 93 is executed by the processor, the dehumidifier online fault diagnosis method provided in this embodiment of the present application is implemented.
The present application provides a computer-readable storage medium, on which a computer program is stored, which, when running on a computer, enables the computer to execute the corresponding content in the foregoing method embodiments.
It should be understood that, although the steps in the flowcharts of the figures are shown in order as indicated by the arrows, the steps are not necessarily performed in order as indicated by the arrows. The steps are not performed in the exact order shown and may be performed in other orders unless explicitly stated herein.
The foregoing is only a few embodiments of the present application and it should be noted that those skilled in the art can make various improvements and modifications without departing from the principle of the present application, and that these improvements and modifications should also be considered as the protection scope of the present application.
Claims (10)
1. An online fault diagnosis method for a dehumidifier is characterized by comprising the following steps:
acquiring point location data representing a point location working state and a fault target point needing to be diagnosed, acquiring other point locations influencing the fault target point according to the fault target point, and acquiring other point location data representing other point location working states based on the other point locations and the point location data;
performing feature extraction operation on the other point location data to obtain a feature value; and processing the characteristic value by using a machine learning model to obtain the prediction data of the fault target point, and judging the running state of the dehumidifier based on the prediction data and the actual data of the fault target point.
2. The method according to claim 1, wherein said obtaining other points affecting the fault target point according to the fault target point comprises:
and judging the fault type of the fault target point according to the fault target point, and determining other point locations influencing the fault target point based on the fault type, wherein each fault type has a judgment standard for determining other point locations influencing the fault target point.
3. The method of claim 1, wherein the feature extraction operation comprises at least one of: and performing operation on one point position data in the other point position data within a period of time, and performing operation on a plurality of point position data in the other point position data within a period of time.
4. The method of claim 3, wherein the processing the eigenvalues using a machine learning model to obtain the predicted data for the fault target point comprises:
acquiring equipment parameters, fault target points and characteristic values of the dehumidifier at regular time, and acquiring historical actual data of the fault target points and historical characteristic values of other points influencing the fault target points according to the fault target points;
determining an initial machine learning model for processing the characteristic value based on the equipment parameters of the dehumidifier and the fault target point, training the initial machine learning model in an off-line manner according to the historical characteristic value and the historical actual data of the fault target point to obtain a final machine learning model, and processing the other point data in an on-line manner by using the final machine learning model to obtain the prediction data of the fault target point.
5. The method of claim 4, wherein the training the initial machine learning model offline according to the historical eigenvalues and the historical actual data of the fault target points further comprises, before obtaining the final machine learning model:
acquiring other point location test data and fault target point test data which are used for testing the accuracy of the final machine learning model in the historical characteristic values and the historical actual data of the fault target points, and processing the other point location test data by using the final machine learning model to acquire test prediction data of the fault target points;
judging whether the difference value between the test prediction data of the fault target point and the actual data of the fault target point is larger than a threshold value, if so, adjusting the feature extraction operation and/or selecting a new initial machine learning model; otherwise, a final machine learning model is obtained.
6. The method of claim 1, wherein the determining the operational state of the dehumidifier based on the predicted data and actual data of a fault target point comprises:
acquiring a deviation between the prediction data and actual data of a fault target point based on the prediction data, recording the number of times that the deviation exceeds a preset deviation within a fixed time, judging whether the number of times exceeds a preset number of times, if so, judging that the running state of the dehumidifier is abnormal, and sending an alarm signal by the dehumidifier; otherwise, the running state of the dehumidifier is normal.
7. An online fault diagnosis system for a dehumidifier, the system comprising: the system comprises a data processing module and an algorithm calculation module; wherein the content of the first and second substances,
the data processing module is used for acquiring point location data representing the working state of the point location;
the algorithm calculation module is used for acquiring a fault target point to be diagnosed, acquiring other point locations influencing the fault target point according to the fault target point, and acquiring other point location data representing the working states of other point locations based on the other point locations and the point location data; performing feature extraction operation on the other point location data to obtain a feature value; and processing the characteristic value by using a machine learning model to obtain the prediction data of the fault target point, and judging the running state of the dehumidifier based on the prediction data and the actual data of the fault target point.
8. The system of claim 7, further comprising a software front-end module, a software back-end module; wherein, the first and the second end of the pipe are connected with each other,
the software front-end module is used for displaying the running state of the dehumidifier and the prediction data of the fault target point;
the software back-end module is used for interacting with the data processing module, the algorithm calculation module and the software front-end module.
9. A server, characterized in that the server comprises: a processor and a memory, the memory having stored thereon a computer program operable on the processor, the computer program, when executed by the processor, implementing a dehumidifier online fault diagnosis method according to any one of claims 1 to 7.
10. A computer-readable storage medium on which a computer program that can be run on a processor is stored, wherein the computer program, when executed by the processor, implements the dehumidifier online fault diagnosis method according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210820907.7A CN115289606A (en) | 2022-07-13 | 2022-07-13 | Dehumidifier online fault diagnosis method, system, server and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210820907.7A CN115289606A (en) | 2022-07-13 | 2022-07-13 | Dehumidifier online fault diagnosis method, system, server and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115289606A true CN115289606A (en) | 2022-11-04 |
Family
ID=83821333
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210820907.7A Pending CN115289606A (en) | 2022-07-13 | 2022-07-13 | Dehumidifier online fault diagnosis method, system, server and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115289606A (en) |
Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109934130A (en) * | 2019-02-28 | 2019-06-25 | 中国空间技术研究院 | The in-orbit real-time fault diagnosis method of satellite failure and system based on deep learning |
| CN110837683A (en) * | 2019-05-20 | 2020-02-25 | 全球能源互联网研究院有限公司 | Training and predicting method and device for prediction model of transient stability of power system |
| CN110991666A (en) * | 2019-11-25 | 2020-04-10 | 远景智能国际私人投资有限公司 | Fault detection method, model training method, device, equipment and storage medium |
| CN111766514A (en) * | 2020-06-19 | 2020-10-13 | 南方电网调峰调频发电有限公司 | Data analysis method for equipment detection points |
| CN111823952A (en) * | 2020-04-17 | 2020-10-27 | 北京嘀嘀无限科技发展有限公司 | Battery cell temperature diagnosis method, storage medium and electronic equipment |
| JP2021018821A (en) * | 2019-07-18 | 2021-02-15 | ▲広▼州大学 | Data intelligent prediction method, device, computer device, and storage medium |
| CN112508053A (en) * | 2020-11-10 | 2021-03-16 | 泽恩科技有限公司 | Intelligent diagnosis method, device, equipment and medium based on integrated learning framework |
| CN112702184A (en) * | 2019-10-23 | 2021-04-23 | 中国电信股份有限公司 | Fault early warning method and device and computer-readable storage medium |
| CN112781178A (en) * | 2021-01-15 | 2021-05-11 | 杭州格米环境科技有限公司 | Dehumidifier online diagnosis system and method |
| CN113011010A (en) * | 2021-03-01 | 2021-06-22 | 常州英集动力科技有限公司 | Boiler fault diagnosis method and diagnosis system based on structural mechanism and operation data |
| KR102285117B1 (en) * | 2021-01-22 | 2021-08-03 | (주)제넥 | Flow control method for absorption type air dryer device using multi dew point meter |
| CN113467420A (en) * | 2021-06-30 | 2021-10-01 | 通号城市轨道交通技术有限公司 | Method and device for detecting zone controller fault |
| WO2021232655A1 (en) * | 2020-05-20 | 2021-11-25 | 国网河北省电力有限公司电力科学研究院 | Method for evaluating mechanical state of high-voltage shunt reactor on the basis of vibration feature |
-
2022
- 2022-07-13 CN CN202210820907.7A patent/CN115289606A/en active Pending
Patent Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109934130A (en) * | 2019-02-28 | 2019-06-25 | 中国空间技术研究院 | The in-orbit real-time fault diagnosis method of satellite failure and system based on deep learning |
| CN110837683A (en) * | 2019-05-20 | 2020-02-25 | 全球能源互联网研究院有限公司 | Training and predicting method and device for prediction model of transient stability of power system |
| JP2021018821A (en) * | 2019-07-18 | 2021-02-15 | ▲広▼州大学 | Data intelligent prediction method, device, computer device, and storage medium |
| CN112702184A (en) * | 2019-10-23 | 2021-04-23 | 中国电信股份有限公司 | Fault early warning method and device and computer-readable storage medium |
| CN110991666A (en) * | 2019-11-25 | 2020-04-10 | 远景智能国际私人投资有限公司 | Fault detection method, model training method, device, equipment and storage medium |
| CN111823952A (en) * | 2020-04-17 | 2020-10-27 | 北京嘀嘀无限科技发展有限公司 | Battery cell temperature diagnosis method, storage medium and electronic equipment |
| WO2021232655A1 (en) * | 2020-05-20 | 2021-11-25 | 国网河北省电力有限公司电力科学研究院 | Method for evaluating mechanical state of high-voltage shunt reactor on the basis of vibration feature |
| CN111766514A (en) * | 2020-06-19 | 2020-10-13 | 南方电网调峰调频发电有限公司 | Data analysis method for equipment detection points |
| CN112508053A (en) * | 2020-11-10 | 2021-03-16 | 泽恩科技有限公司 | Intelligent diagnosis method, device, equipment and medium based on integrated learning framework |
| CN112781178A (en) * | 2021-01-15 | 2021-05-11 | 杭州格米环境科技有限公司 | Dehumidifier online diagnosis system and method |
| KR102285117B1 (en) * | 2021-01-22 | 2021-08-03 | (주)제넥 | Flow control method for absorption type air dryer device using multi dew point meter |
| CN113011010A (en) * | 2021-03-01 | 2021-06-22 | 常州英集动力科技有限公司 | Boiler fault diagnosis method and diagnosis system based on structural mechanism and operation data |
| CN113467420A (en) * | 2021-06-30 | 2021-10-01 | 通号城市轨道交通技术有限公司 | Method and device for detecting zone controller fault |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| JP7069269B2 (en) | Semi-supervised methods and systems for deep anomaly detection for large industrial surveillance systems based on time series data using digital twin simulation data | |
| CN110320892B (en) | Sewage treatment equipment fault diagnosis system and method based on L asso regression | |
| CN109240244B (en) | Data-driven equipment running state health degree analysis method and system | |
| CN111047732B (en) | Equipment abnormity diagnosis method and device based on energy consumption model and data interaction | |
| RU2313815C2 (en) | Device and method for controlling technical plant, which contains a set of systems, in particular, electric power plant | |
| CN107291991B (en) | Early defect early warning method for wind turbine generator based on dynamic network sign | |
| CN112799898B (en) | Interconnection system fault node positioning method and system based on distributed fault detection | |
| US20120271826A1 (en) | Data collecting method for detection and on-time warning system of industrial process | |
| US11953894B2 (en) | Building management system with machine learning for detecting anomalies in vibration data sets | |
| CN112836941B (en) | Online health condition assessment method for high-pressure system of steam turbine of coal motor unit | |
| CN105242660A (en) | Multi-modal cigarette primary processing process online monitoring and fault diagnosis method based on relative change analysis | |
| US20220244682A1 (en) | Edge devices and gateways with machine learning for detecting anomalies in building equipment vibration data | |
| CN117032120A (en) | Integrated intelligent cloud control system and control method for air compression station | |
| CN111340307B (en) | Method for predicting wind power generation power of fan and related device | |
| CN112070353A (en) | Method and system for accurately detecting energy efficiency of data center | |
| CN115575579A (en) | Carbon monitoring method and system based on monitoring source analysis | |
| CN112132316A (en) | System and method for monitoring abnormality of on-line equipment in silk making link | |
| CN114970101A (en) | Management method and system of dehumidification device, electronic equipment and computer storage medium | |
| JP2005301582A (en) | Process management device | |
| CN111275331A (en) | Monitoring method and device for production system | |
| CN114648212A (en) | Cloud computing-based ship equipment performance intelligent management system and method | |
| CN113390641A (en) | Intelligent early warning and online diagnosis method and system for equipment faults of wind and smoke system | |
| CN117091273A (en) | Control method and device for self-adaptive learning data model of central air conditioning system | |
| CN115289606A (en) | Dehumidifier online fault diagnosis method, system, server and storage medium | |
| CN110108457B (en) | Primary fan shaft temperature diagnosis method based on universal gravitation neural network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |