CN115526274A - Method, system, equipment and medium for diagnosing faults of sensors in air handling unit - Google Patents
Method, system, equipment and medium for diagnosing faults of sensors in air handling unit Download PDFInfo
- Publication number
- CN115526274A CN115526274A CN202211340653.5A CN202211340653A CN115526274A CN 115526274 A CN115526274 A CN 115526274A CN 202211340653 A CN202211340653 A CN 202211340653A CN 115526274 A CN115526274 A CN 115526274A
- Authority
- CN
- China
- Prior art keywords
- sensor
- handling unit
- air handling
- data
- fault
- 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
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Mathematical Physics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Computational Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Optimization (AREA)
- Mathematical Analysis (AREA)
- Software Systems (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- Biomedical Technology (AREA)
- Artificial Intelligence (AREA)
- Computational Linguistics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- Biophysics (AREA)
- Algebra (AREA)
- Databases & Information Systems (AREA)
- Testing And Monitoring For Control Systems (AREA)
Abstract
The invention discloses a method, a system, equipment and a medium for diagnosing the fault of a sensor in an air handling unit, wherein the method comprises the following steps: acquiring real-time data of a sensor in an air handling unit to be predicted; taking the real-time data of the sensor in the air handling unit to be predicted as the input of a pre-constructed micro fault diagnosis model of the sensor of the air handling unit, and outputting to obtain the fault diagnosis result of the sensor in the air handling unit to be predicted; optimizing the kernel parameters of the kernel principal component correlation analysis algorithm by using a particle swarm optimization algorithm to obtain an optimized kernel principal component correlation analysis algorithm; replacing Euclidean distances in a kernel principal component analysis algorithm by using the correlation distances to optimize a Gaussian radial basis kernel function in the kernel principal component analysis algorithm to obtain the kernel principal component correlation analysis algorithm; the invention can diagnose and detect the tiny faults of the sensors in the air handling unit, obtain the fault area and the fault sensor and has higher fault detection efficiency.
Description
Technical Field
The invention belongs to the technical field of air conditioning systems, and particularly relates to a method, a system, equipment and a medium for diagnosing faults of a sensor in an air handling unit.
Background
Heating Ventilation and Air Conditioning (HVAC) systems are used as main equipment for guaranteeing the thermal comfort of indoor human bodies, the energy consumption amount of the HVAC systems accounts for about 50% of the total energy consumption in the operation stage of buildings, and the Air Conditioning energy consumption in commercial buildings and comprehensive buildings can reach more than 65%; therefore, energy utilization of HVAC systems is critical to saving building energy consumption; to ameliorate this problem, a variety of optimization strategies have been applied to HVAC systems; for example: optimization, monitoring, automatic diagnosis, and the like. Research has shown that by using fault detection and diagnostic techniques in commercial buildings, energy consumption of HVAC systems can be reduced by 20% -30%.
The sensor is used as a signal source for realizing intelligent control of the HVAC system and is a component widely used in the HVAC system, such as a temperature sensor, a humidity sensor, a pressure sensor, a flow sensor and the like; therefore, ensuring the accuracy of the measured data of the sensors is vital to realizing the intelligent control of the building energy system; however, during operation of the HVAC system, sensor failure is inevitable; when a sensor for control fails, the control system may issue an erroneous command, causing the actuator to perform an incorrect operation; while such failures can be addressed by manually adjusting the control strategy, as sensor failures continue to deepen, they will result in increased HVAC system energy consumption costs; meanwhile, as the structural complexity of HVAC systems of commercial buildings gradually increases, the maintenance cost of sensors also increases.
Therefore, the sensor fault diagnosis research based on the heating and ventilation system has important significance for building energy conservation and environmental protection; air conditioning systems are generally composed of an air system and a water system; wherein, the wind system (namely the air handling unit) is the core component in the air conditioning system; sensors in the air handling unit are in a high-temperature working environment for a long time, the sensors are prone to failure, and tiny failures with low failure degrees are prone to occurring; relatively speaking, sensor failure occurs not at one stroke, which has a potential, gradual process; therefore, if the fault signal is captured in time at the initial stage of the sensor fault, the influence of the fault on the system operation is greatly reduced, and therefore, the detection research of the minor fault is very necessary.
The air handling unit is used as a key component of an air conditioning system, wherein a complex control system is arranged, and the indoor comfort level is reduced and even equipment is damaged due to the failure of a sensor of the air handling unit; at present, most of the existing sensor fault diagnosis methods are limited to sensor faults with higher fault degree, and the diagnosis and detection methods for micro faults of sensors with lower fault degree are not sufficient, and because the micro fault characteristics are not obvious, the fault characteristics are still difficult to extract at the initial stage of the fault; therefore, it is desirable to provide a method for diagnosing minor faults of a sensor of an air handling unit.
Disclosure of Invention
Aiming at the technical problems in the prior art, the invention provides a method, a system, equipment and a medium for diagnosing the faults of sensors in an air handling unit, which are used for solving the technical problems that the existing sensor fault diagnosis method is mostly limited to the faults of the sensors with higher fault degree and can not meet the diagnosis and detection of the tiny faults of the sensors with lower fault degree.
In order to achieve the purpose, the invention adopts the technical scheme that:
the invention provides a method for diagnosing faults of a sensor in an air handling unit, which comprises the following steps:
acquiring real-time data of a sensor in an air handling unit to be predicted;
taking the real-time data of the sensor in the air handling unit to be predicted as the input of a pre-constructed micro fault diagnosis model of the sensor of the air handling unit, and outputting to obtain the fault diagnosis result of the sensor in the air handling unit to be predicted;
the method comprises the following steps of constructing a pre-constructed micro fault diagnosis model of the air handling unit sensor, wherein the pre-constructed micro fault diagnosis model of the air handling unit sensor comprises the following steps:
optimizing the kernel parameters of the kernel principal component correlation analysis algorithm by using a particle swarm optimization algorithm to obtain an optimized kernel principal component correlation analysis algorithm; replacing Euclidean distances in a kernel principal component analysis algorithm by using correlation distances to optimize Gaussian radial basis kernel functions in the kernel principal component analysis algorithm to obtain the kernel principal component correlation analysis algorithm;
training the optimized kernel principal component correlation analysis algorithm by using sensor historical data of a normal air handling unit to obtain a pre-constructed micro fault diagnosis model of the air handling unit sensor; and the historical data of the sensor of the normal air handling unit comprises normal data and preset fault data.
Further, the normal data comprises historical data of sensors in the normal air handling unit; the sensor historical data in the normal air handling unit comprises the opening degree of a chilled water valve, the temperature of fresh air, the humidity of the fresh air, the temperature of supplied air, the humidity of supplied air, the temperature of returned air and the humidity of returned air.
Furthermore, the preset fault data is acquired after a fault form is preset on a sensor in a normal air handling unit; wherein the preset fault forms comprise 5% -20% of drift fault forms or 5% -20% of deviation fault forms.
Further, the process of obtaining the fault diagnosis result of the sensor in the air handling unit to be predicted by outputting the real-time data of the sensor in the air handling unit to be predicted as the input of the pre-constructed micro fault diagnosis model of the sensor in the air handling unit to be predicted specifically comprises the following steps:
according to the real-time data of the sensor in the air handling unit to be predicted, an initial data matrix X is constructed and obtained N×m ；
Combining the principle that the information entropy is unchanged before and after linear change, and aligning the initial data matrix X N×m Linear transformation is carried out to obtain a homogeneous data matrix Z N×m ；
Use the instituteThe optimized kernel principal component correlation analysis algorithm is respectively applied to the initial data matrix X N×m And the homogeneous data matrix Z N×m Performing dimensionality reduction to obtain an initial data dimensionality reduction result Y N×m And homogenous data dimensionality reduction result Y' N×m ；
Calculating the dimensionality reduction result Y of the initial data N×m And homogenous data dimensionality reduction result Y' N×m Relative distance D between i ；
Correlating the distance D i With a predetermined minimum correlation distance D min Comparing if the related distance D i Less than a predetermined minimum correlation distance D min And extracting fault characteristics by using the principal component contribution rate, and outputting to obtain a sensor fault diagnosis result in the air handling unit to be predicted.
Further, the initial data matrix X N×m Comprises the following steps:
wherein N is the number of the real-time data of the sensors, and m is the number of the sensors; x' Nm Real-time data of an Nth sensor of the mth sensor;
the homogeneous data matrix Z N×m Comprises the following steps:
wherein Z is ij ' is the result of linear transformation of the ith sensor real-time data of the jth sensor; x is the number of i ′ j Real-time data of an ith sensor of the jth sensor; x is the number of i ′ (j+1) The ith sensor real-time data of the (j + 1) th sensor; x is the number of i ′ m Is as followsThe ith sensor real-time data of the m sensors; x is a radical of a fluorine atom i ′ 1 Real-time data of the ith sensor of the 1 st sensor.
Further, if the correlation distance D i Greater than or equal to a preset minimum correlation distance D min Updating the kernel parameters of the kernel principal component correlation analysis algorithm by using a particle swarm algorithm to obtain an updated kernel principal component correlation analysis algorithm;
respectively carrying out correlation analysis on the initial data matrix X by using the updated kernel principal component correlation analysis algorithm N×m And the homogeneous data matrix Z N×m And carrying out dimension reduction processing and calculating the related distance again.
Further, when the principal component contribution rate is used to extract the fault feature, the contribution rate CPV (i) of the ith principal component is:
wherein λ is i The characteristic value of the ith principal element, namely the ith kernel function, of the principal component analysis method is obtained; n is the total number of principal elements of the principal component analysis method;
the cumulative contribution CPV of the first p principal elements is:
the invention also provides a system for diagnosing the fault of the sensor in the air handling unit, which comprises the following components:
the data acquisition module is used for acquiring the real-time data of the sensor in the air handling unit to be predicted;
the diagnosis output module is used for taking the real-time data of the sensor in the air handling unit to be predicted as the input of a pre-constructed micro fault diagnosis model of the sensor of the air handling unit and outputting to obtain the fault diagnosis result of the sensor in the air handling unit to be predicted;
the method comprises the following steps of constructing a pre-constructed micro fault diagnosis model of the air handling unit sensor, wherein the pre-constructed micro fault diagnosis model of the air handling unit sensor comprises the following steps:
optimizing the kernel parameters of the kernel principal component correlation analysis algorithm by using a particle swarm algorithm to obtain an optimized kernel principal component correlation analysis algorithm; replacing Euclidean distances in a kernel principal component analysis algorithm by using correlation distances to optimize Gaussian radial basis kernel functions in the kernel principal component analysis algorithm to obtain the kernel principal component correlation analysis algorithm;
training the optimized kernel principal component correlation analysis algorithm by using sensor historical data of a normal air handling unit to obtain a pre-constructed micro fault diagnosis model of the air handling unit sensor; and the historical data of the sensor of the normal air handling unit comprises normal data and preset fault data.
The invention also provides a sensor fault diagnosis device in the air handling unit, which comprises:
a memory for storing a computer program;
and the processor is used for realizing the steps of the method for diagnosing the fault of the sensor in the air handling unit when executing the computer program.
The invention also provides a computer-readable storage medium, which stores a computer program that, when executed by a processor, implements the steps of the method for diagnosing a sensor fault in an air handling unit.
Compared with the prior art, the invention has the beneficial effects that:
the invention provides a method and a system for diagnosing the fault of a sensor in an air handling unit, which realize the simulation of each fault state possibly occurring in the air handling unit by adding preset fault data into the historical data of the sensor of a normal air handling unit; the method comprises the steps of optimizing nuclear parameters in a KPCCA method by using a PSO algorithm, and then extracting characteristics of sensor data by using the optimized KPCCA algorithm to obtain a specific fault type, thereby effectively solving the problem that the tiny faults of the sensor are difficult to identify in the diagnosis process, effectively improving the tiny fault diagnosis efficiency of the sensor in the air handling unit and improving the diagnosis success rate; the invention does not need to artificially identify the suspicious target, and realizes automatic fault detection through the integral operation state of the data; the method comprises the following steps of (1) diagnosing and detecting tiny faults of sensors in the air handling unit, and acquiring a fault area and a fault sensor; the fault detection efficiency is high, the tiny fault type can be accurately obtained, and the blank of tiny fault diagnosis of the sensor in the air handling unit is effectively filled.
Drawings
FIG. 1 is a flow chart of a method of diagnosing a fault in a sensor of an air handling unit according to the present invention;
FIG. 2 is a flow chart of a particle swarm algorithm in the present invention;
fig. 3 is a diagram of detection results of PSO-KPCCA in different drift faults in the prior KPCA and the embodiment;
FIG. 4 is a diagram of classification of 5% sensor drift fault data for a test sample for a conventional KPCA and a PSO-KPCCA in an embodiment;
FIG. 5 is a diagram of a conventional KPCA and the classification of 10% sensor drift fault data of a PSO-KPCCA in an embodiment on a test sample;
FIG. 6 is a diagram of classification of 15% sensor drift fault data for a test sample for a PSO-KPCCA in a conventional KPCA and an embodiment;
FIG. 7 is a diagram of classification of 20% sensor drift fault data for a test sample for a PSO-KPCCA in a conventional KPCA and an embodiment;
fig. 8 is a diagram of detection results of a conventional KPCA and a PSO-KPCCA in an embodiment under bias faults of different degrees;
FIG. 9 is a diagram of a conventional KPCA and an embodiment of a PSO-KPCCA for classifying 5% sensor offset failure data of a test sample;
FIG. 10 is a diagram of a conventional KPCA and the classification of 10% sensor offset failure data of a PSO-KPCCA in an embodiment on a test sample;
FIG. 11 is a diagram of 15% sensor offset fault data classification for a test sample for a PSO-KPCCA in a conventional KPCA and an embodiment;
fig. 12 is a diagram of classification of 20% sensor offset failure data of a test sample for a PSO-KPCCA in an embodiment and a conventional KPCA.
Detailed Description
In order to make the technical problems, technical solutions and advantageous effects of the present invention more apparent, the following embodiments further describe the present invention in detail. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
As shown in the attached figure 1, the invention provides a method for diagnosing the fault of a sensor in an air handling unit, which comprises the following steps:
Step 2, constructing a micro fault diagnosis model of the air handling unit sensor to obtain a pre-constructed micro fault diagnosis model of the air handling unit sensor; the construction process specifically comprises the following steps:
optimizing the kernel parameters of the kernel principal component correlation analysis algorithm by using a particle swarm algorithm to obtain an optimized kernel principal component correlation analysis algorithm; replacing Euclidean distances in a kernel principal component analysis algorithm by using the correlation distances so as to optimize a Gaussian radial basis kernel function in the kernel principal component analysis algorithm to obtain the kernel principal component correlation analysis algorithm;
and training the optimized nuclear principal component correlation analysis algorithm by using the sensor historical data of the normal air handling unit to obtain the pre-constructed air handling unit sensor tiny fault diagnosis model.
In the invention, the historical data of the sensor of the normal air handling unit comprises normal data and preset fault data; the normal data comprise historical data of sensors in the normal air processing unit; the historical data of the sensors in the normal air handling unit comprises the opening degree of a chilled water valve, the temperature of fresh air, the humidity of the fresh air, the temperature of supplied air, the humidity of supplied air, the temperature of returned air and the humidity of returned air; the preset fault data is acquired by presetting a fault form for a sensor in a normal air handling unit; wherein the preset fault patterns comprise 5% -20% of drift fault patterns or 5% -20% of deviation fault patterns.
Step 3, taking the real-time data of the sensor in the air handling unit to be predicted as the input of a pre-constructed sensing tiny fault diagnosis model of the air handling unit, and outputting to obtain the fault diagnosis result of the sensor in the air handling unit to be predicted;
the specific process is as follows:
step 31, obtaining an initial data matrix X according to the real-time data of the sensor in the air handling unit to be predicted N×m ；
Wherein the initial data matrix X N×m Is composed of
Wherein N is the number of the real-time data of the sensors, and m is the number of the sensors; x' Nm Real-time data of an Nth sensor of the mth sensor.
Step 32, combining the principle that the information entropy is unchanged before and after linear change, and aligning the initial data matrix X N×m Linear transformation is carried out to obtain a homogeneous data matrix Z N×m (ii) a Specifically, the principle that the information entropy is unchanged before and after linear change is combined, and the initial data matrix X is subjected to N×m Carrying out linear transformation to obtain a matrix containing original data information, namely obtaining the homogeneous data matrix Z N×m ；
Wherein the homogeneous data matrix is Z N×m ：
Wherein, Z ij ' is the result of linear transformation of the ith sensor real-time data of the jth sensor; x is a radical of a fluorine atom i ′ j Real-time data of an ith sensor of the jth sensor; x is the number of i ′ (j+1) The ith sensor real-time data of the (j + 1) th sensor; x is the number of i ′ m Real-time data of an ith sensor of the mth sensor; x is the number of i ′ 1 Real-time data of the ith sensor of the 1 st sensor.
Step 33, using the optimized kernel principal component correlation analysis algorithm to respectively perform the initial data matrix X N×m And the homogeneous data matrix Z N×m Performing dimensionality reduction to obtain an initial data dimensionality reduction result Y N×m And homogenous data dimensionality reduction result Y' N×m 。
Step 34, calculating the dimensionality reduction result Y of the initial data N×m And homogenous data dimensionality reduction result Y' N×m Relative distance D therebetween i 。
Step 35, correlating the distance D i With a predetermined minimum correlation distance D min Comparing if the related distance D i Less than a predetermined minimum correlation distance D min Extracting fault characteristics by using principal component contribution rate, and outputting to obtain a sensor fault diagnosis result in the air handling unit to be predicted;
when the principal component contribution rate is used for extracting the fault features, the contribution rate CPV (i) of the ith principal component is as follows:
wherein λ is i The characteristic value of the ith principal element, namely the ith kernel function, of the principal component analysis method is obtained; n is the total number of principal elements of the principal component analysis method;
the cumulative contribution CPV of the first p principal elements is:
step 36, if the related distance D i Greater than or equal to a preset minimum correlation distance D min Updating the kernel parameters of the kernel principal component correlation analysis algorithm by using a particle swarm algorithm; obtaining an updated kernel principal component correlation analysis algorithm, returning to the step 33-34, and performing the operation of dimension reduction processing and calculating the correlation distance again; wherein, a particle swarm algorithm is used to update the kernel parameters of the kernel principal component correlation analysis algorithm, as shown in fig. 2.
According to the sensor fault diagnosis direction in the air handling unit, on the basis of a kernel principal component analysis method (KPCA), according to the principle that the value of information entropy is not changed, the traditional Euclidean distance is replaced by the related distance to optimize parameters of a Gaussian radial basis kernel function, the kernel principal component correlation analysis method (KPCCA) is provided for fault feature extraction and diagnosis, the kernel parameters in the KPCCA algorithm are optimized through a Particle Swarm Optimization (PSO), and therefore the feature extraction of tiny sensor faults is completed so as to diagnose.
The invention also provides a system for diagnosing the fault of the sensor in the air handling unit, which comprises a data acquisition module and a diagnosis output module; the data acquisition module is used for acquiring the real-time data of the sensor in the air handling unit to be predicted; and the diagnosis output module is used for outputting the real-time data of the sensor in the air handling unit to be predicted as the input of a pre-constructed micro fault diagnosis model of the sensor of the air handling unit to obtain the fault diagnosis result of the sensor in the air handling unit to be predicted.
The invention also provides a device for diagnosing the fault of the sensor in the air handling unit, which comprises: a memory for storing a computer program; and the processor is used for realizing the steps of the method for diagnosing the fault of the sensor in the air handling unit when the computer program is executed.
The processor, when executing the computer program, implements the steps of the method for diagnosing a sensor fault in an air handling unit, for example: acquiring real-time data of a sensor in an air handling unit to be predicted; and taking the real-time data of the sensor in the air handling unit to be predicted as the input of a pre-constructed micro fault diagnosis model of the sensor of the air handling unit, and outputting to obtain the fault diagnosis result of the sensor in the air handling unit to be predicted.
Alternatively, the processor implements the functions of the modules in the system when executing the computer program, for example: the data acquisition module is used for acquiring the real-time data of the sensor in the air handling unit to be predicted; and the diagnosis output module is used for taking the real-time data of the sensor in the air handling unit to be predicted as the input of a pre-constructed micro fault diagnosis model of the sensor of the air handling unit and outputting to obtain the fault diagnosis result of the sensor in the air handling unit to be predicted.
Illustratively, the computer program may be partitioned into one or more modules/units, stored in the memory and executed by the processor, to implement the invention. The one or more modules/units may be a series of computer program instruction segments capable of performing preset functions, the instruction segments being used to describe the execution of the computer program in the sensor fault diagnosis device within the air handling unit. For example, the computer program may be divided into a data acquisition module and a diagnostic output module; the specific functions of each module are as follows: the data acquisition module is used for acquiring real-time data of a sensor in the air handling unit to be predicted; and the diagnosis output module is used for outputting the real-time data of the sensor in the air handling unit to be predicted as the input of a pre-constructed micro fault diagnosis model of the sensor of the air handling unit to obtain the fault diagnosis result of the sensor in the air handling unit to be predicted.
The sensor fault diagnosis device in the air handling unit can be a desktop computer, a notebook computer, a palm computer, a cloud server and other computing devices. The sensor fault diagnosis device in the air handling unit can comprise a processor and a memory, but is not limited to the processor and the memory. Those skilled in the art will appreciate that the above is exemplary of an in-air handling unit sensor fault diagnostic device, and does not constitute a limitation of an in-air handling unit sensor fault diagnostic device, and may include more components than those described above, or some components in combination, or different components, for example, the in-air handling unit sensor fault diagnostic device may also include input output devices, network access devices, buses, etc.
The processor may be a Central Processing Unit (CPU), other general purpose processor, a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), an off-the-shelf programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic, discrete hardware components, etc. The general purpose processor may be a microprocessor or the processor may be any conventional processor or the like, the processor being the control center for the sensor fault diagnosis device in the air handling unit, with various interfaces and lines connecting the various parts of the sensor fault diagnosis device throughout the air handling unit.
The memory may be used to store the computer programs and/or modules, and the processor may implement the various functions of the sensor fault diagnosis device in the air handling unit by executing or executing the computer programs and/or modules stored in the memory and invoking data stored in the memory.
The memory may mainly include a storage program area and a storage data area, wherein the storage program area may store an operating system, an application program required by at least one function (such as a sound playing function, an image playing function, etc.), and the like; the storage data area may store data (such as audio data, a phonebook, etc.) created according to the use of the cellular phone, etc. In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) card, a flash memory card (FlashCard), at least one magnetic disk storage device, a flash memory device, or other volatile solid state storage device.
The invention also provides a computer-readable storage medium storing a computer program which, when executed by a processor, implements the steps of a method for diagnosing a fault in a sensor of an air handling unit.
The modules/units integrated with the sensor fault diagnosis system in the air handling unit may be stored in a computer readable storage medium if implemented as software functional units and sold or used as separate products.
Based on such understanding, the present invention may implement all or part of the processes of the method for diagnosing the sensor fault in the air handling unit, and may also be implemented by using a computer program to instruct related hardware, where the computer program may be stored in a computer readable storage medium, and when the computer program is executed by a processor, the steps of the method for diagnosing the sensor fault in the air handling unit may be implemented. Wherein the computer program comprises computer program code, which may be in source code form, object code form, executable file or preset intermediate form, etc.
The computer-readable storage medium may include: any entity or device capable of carrying the computer program code, recording medium, usb disk, removable hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random Access Memory (RAM), electrical carrier wave signals, telecommunications signals, software distribution medium, etc.
It should be noted that the computer readable storage medium may contain content that is subject to appropriate increase or decrease as required by legislation and patent practice in jurisdictions, for example, in some jurisdictions, computer readable storage media that does not include electrical carrier signals and telecommunications signals in accordance with legislation and patent practice.
Examples
The embodiment provides a fault diagnosis method for a sensor in an air handling unit, which comprises the following steps:
In general, the sensitivity of each HVAC system characteristic parameter to different categories and different fault levels under different operating conditions is different, and the correlation between different characteristic parameters is also different under different conditions; therefore, the method for analyzing the nuclear principal component is adopted to extract the fault characteristics of the sensors in the air handling unit, so that the correlation among the characteristics can be reduced, the redundancy is reduced, the sensitivity is improved, and the dimensionality is reduced.
Assuming that the subject contains m sensors, the sensor data set for the t-th measurement is represented as:
wherein, Y (t) A sensor history data set for the t-th measurement;
historical data of the sensor measured t time for the mth sensor.
When the historical data measured by the m sensors for N times are expressed in a matrix form, the historical data matrix of the sensors measured by all the sensors for N times is as follows:
y is a sensor historical data matrix measured by all sensors for N times; y is Nm Sensor history data measured for the nth time of the mth sensor; thus, N samples of measured values under normal sensor operating conditions can be constructed as Y ∈ R N×m Of the matrix of (a).
Step 2, standardizing the sensor historical data of the normal air handling unit according to a standardized processing formula to eliminate the influence of different sensor dimensions on fault feature extraction and obtain a standardized matrix of historical sampling data;
wherein the normalization processing formula is:
wherein x is ij Normalizing the historical data of the sensor measured by the jth sensor at the ith time; y is ij Sensor history data measured for the ith time of the jth sensor;
is the average of the sensor history data of the jth sensor;
is the standard deviation of the sensor history data for the jth sensor.
Wherein the normalized matrix of the historical sample data is:
wherein X is a standardized matrix of the historical sampling data.
Step 3, combining a principle that information entropy is unchanged before and after linear change, and performing linear transformation on the standardized matrix X of the historical sampling data to obtain a historical homogeneous data matrix Z; specifically, a principle that information entropy is unchanged before and after linear change is combined, linear transformation is carried out on the standardized matrix X of the historical sampling data, a matrix containing original historical data information is obtained, and the historical homogeneous data matrix Z is obtained;
wherein the historical homogenous data matrix is Z:
wherein Z is ij The method is a result of linear transformation of ith sensor historical data of a jth sensor; x is the number of ij Ith sensor history data for a jth sensor; x is the number of i(j+1) Ith sensor history data for the (j + 1) th sensor; x is the number of im Ith sensor history data for the mth sensor; x is the number of i1 Is the ith sensor history data for the 1 st sensor.
Step 4, optimizing the kernel parameters of the Gaussian radial basis kernel function by combining a correlation method and utilizing a particle swarm algorithm to obtain optimized kernel parameters; the specific process is as follows:
setting a nuclear parameter σ i Is within the empirical range [ sigma ] min ,σ max ]；
Using particle swarm optimization to correct the kernel parameter sigma i Updating is carried out;
taking the updated kernel parameter as the kernel parameter of the Gaussian radial basis kernel function, and respectively performing dimensionality reduction on the standardized matrix X of the historical sampling data and the historical homogeneous data matrix Z by using a KPCA (kernel principal component analysis) method to obtain a dimensionality reduction result Y of the historical sampling data N×r And historical homogenous data dimensionality reduction result Y' N×r ；
Calculating the dimensionality reduction result Y of the initial data N×m And homogenous data dimension reduction result Y' N×m Relative distance D between i So as to measure the dimensionality reduction knotDeviation of fruit; wherein the correlation distance D is calculated according to the following formula i ；
D i ＝1-ρ i
When the correlation distance D i Less than a predetermined minimum correlation distance D min Then take the minimum correlation distance D i The corresponding nuclear parameter is the optimal nuclear parameter, and the optimized nuclear parameter is obtained; otherwise i = i +1, and the operation of step 4 is repeated.
wherein sigma is the optimized nuclear parameter.
Step 6, carrying out normalization processing on the Gaussian radial basis kernel function by using the following formula to obtain a normalized kernel function;
wherein the content of the first and second substances,
is a normalized gaussian radial basis kernel function.
Step 7, calculating a characteristic value and a characteristic vector of the normalized Gaussian radial basis function by using the following formula;
wherein, λ is the characteristic value of the normalized Gaussian radial basis kernel function;
is the feature vector of the normalized Gaussian radial basis kernel function.
Step 8, performing normalization processing on the feature connection by using the following formula;
step 9, determining the number of the principal elements by using the principal element contribution rate
Research finds that if the number of the selected pivot elements is too large, the residual subspace contains less information; if the number of the pivot elements is too small, information redundancy will be caused, and effective information extraction cannot be realized. Therefore, the number of pivot elements will directly affect the extraction of features; in this embodiment, the fault feature is extracted based on a pivot probability (CPV).
When the principal component contribution rate is used for extracting the fault features, the contribution rate CPV (i) of the ith principal component is as follows:
wherein λ is i The characteristic value of the ith principal element, namely the ith kernel function, of the principal component analysis method is obtained; n is the total number of principal elements of the principal component analysis method;
the cumulative contribution CPV of the first p principal elements is:
the contribution rate of the accumulated variance reflects the representation capability of the selected principal element on the whole data, and if the CPV value is too small, the selected principal element contains less information on the whole, so that the problems that the research based on the selected principal element is not accurate enough, the related content cannot be reflected truly and the like are caused; if the CPV value is too large, the information covered by the selected pivot element is too numerous and complicated, and contains a large amount of secondary or useless information, so that the primary information is ignored; the CPV value was set to 90% in combination with the previous experience.
And selecting the characteristic vectors corresponding to the previous p rows from the homogeneous data matrix, and constructing the matrix after unitization to obtain a reduced-dimension sample, so that the extraction of the fault characteristics is realized, a trained PSO-KPCCA fault diagnosis model is obtained, and the tiny fault diagnosis model of the air handling unit sensor is obtained.
And 11, acquiring real-time data of a sensor in the air handling unit to be predicted.
And 12, taking the real-time data of the sensor in the air handling unit to be predicted as the input of a pre-constructed sensing tiny fault diagnosis model of the air handling unit, and outputting to obtain the fault diagnosis result of the sensor in the air handling unit to be predicted.
Description of specific tests:
taking the fault diagnosis process of an air handling unit of a certain air conditioning system as an example, the collected sensor data is summer working condition data.
In the embodiment, the total collected sensor data of the normal water chilling unit at different time periods is 1000 groups, including normal data and eight kinds of preset fault data; wherein, under each preset fault form, 300 groups of sensor data are collected; the data are processed by using the existing KPCA method verified by the euro method and the method for diagnosing the fault of the sensor in the air handling unit (PSO-KPCCA) described in this embodiment, so as to obtain a diagnosis result by using the existing KPCA method and a diagnosis result by using the method for diagnosing the fault of the sensor in the air handling unit described in this embodiment, which are specifically as follows:
by using historical data of sensors of a normal air processing unit, fault feature extraction and diagnosis are mainly performed by a kernel principal component correlation analysis method (KPCCA) in the embodiment, kernel parameters in a KPCCA algorithm are optimized by a Particle Swarm Optimization (PSO), and therefore fault diagnosis of a tiny sensor is completed; arranging normal sensors or fault sensors according to a preset fault form at sensor point positions of a normal air handling unit at different time intervals; the preset fault modes comprise 5%, 10%, 15% and 20% drift and deviation faults; and collecting data of normal sensors or fault sensors at different time intervals to obtain real-time sensor data of the air handling unit.
In this embodiment, eight preset fault modes are set; under normal conditions and in different time periods with eight preset fault forms, the sensor data are used as experiments to acquire sensor parameters in the air conditioner cold-heat exchanger in real time, and the sensor real-time data of the normal air handling unit are acquired.
And the acquired opening sensor information of the chilled water valve, the fresh air temperature sensor information, the fresh air humidity sensor information, the air supply temperature sensor information, the return air temperature sensor information and the return air humidity sensor information are acquired. Then, according to a time sequence, constructing a sensor data matrix by using the sensor data at the same moment in the sample, and carrying out normalization processing; the mathematical expression of the sensor data matrix is as follows: dividing the sensor data matrix subjected to normalization processing into two groups according to a preset fault data form and a time sequence to obtain a training data group and a test data group; each training data set and each testing data set comprise normal data and various preset fault data; using the training data group as a training sample of the PSO-KPCCA model to obtain a trained fault diagnosis model; the trained fault diagnosis model is a PSO-KPCCA fault feature extraction and diagnosis model with trained parameters.
And (3) taking the test data group as a test sample of the trained PSO-KPCCA fault feature extraction and diagnosis model, evaluating and adjusting parameters of the trained PSO-KPCCA fault feature extraction and diagnosis model, and stopping iteration after finding an optimal check function value to obtain the PSO-KPCCA model with optimal parameters, namely the fault feature extraction and diagnosis model.
As can be seen from fig. 3 to 7, the method for diagnosing the fault of the sensor in the air handling unit has good diagnostic capability, and the accuracy of the fault diagnosis under the four-degree drifting fault grouping is 65.67%, 67.67%, 68.67% and 70% respectively; compared with the existing PSO-KPCCA method which adopts Euclidean distance to optimize nuclear parameters, the accuracy of the method is respectively improved by 10.67%, 11.00%, 9.67% and 9.67% compared with the accuracy of the related method of the Euclidean KPCA under the drift faults of 5%, 10%, 15% and 20%; the success rate of the drift fault diagnosis comprehensively reaches 68.1%, and the accuracy of the diagnosis result of the minor fault is effectively improved.
As can be seen from fig. 8 to 12, the method for diagnosing the fault of the sensor in the air handling unit has good diagnostic capability, and the accuracy of the fault diagnosis under the deviation fault grouping of four degrees is 99.33%; compared with the existing KPCA method for optimizing the nuclear parameters by adopting the Euclidean distance, the fault diagnosis accuracy is improved by about 2% under the condition of 5% deviation fault degree, and the accuracy is close to that under other fault degrees.
For a description of relevant parts in the system, the device, and the computer-readable storage medium for diagnosing the fault of the sensor in the air handling unit provided in this embodiment, reference may be made to detailed descriptions of corresponding parts in the method for diagnosing the fault of the sensor in the air handling unit described in this embodiment, and details are not repeated here.
According to the invention, the problem that the diagnosis of the sensor fault of the existing air processing system is limited to the fault detection by using a neural network or a clustering method is avoided; in the fault diagnosis process, a suspicious target does not need to be artificially identified and detected, and fault detection can be automatically carried out according to the integral operation state of sensor data; the invention can meet the requirements of extracting the characteristics of the sensor faults with lower degree and diagnosing the sensor faults, realizes the detection of the fault conditions of the micro sensors and finds out the fault areas and the faulty sensors; compared with the KPCA algorithm optimized by using the traditional Euclidean method, the PSO-KPCCA algorithm provided by the invention improves the fault detection rate, and adopts the feature extraction concept to greatly reduce the confusion probability of micro fault diagnosis which is difficult to distinguish.
The above-described embodiment is only one of the embodiments that can implement the technical solution of the present invention, and the scope of the present invention is not limited by the embodiment, but includes any variations, substitutions and other embodiments that can be easily conceived by those skilled in the art within the technical scope of the present invention disclosed.
Claims (10)
1. A method for diagnosing faults of sensors in an air handling unit is characterized by comprising the following steps:
acquiring real-time data of a sensor in an air handling unit to be predicted;
taking the real-time data of the sensor in the air handling unit to be predicted as the input of a pre-constructed micro fault diagnosis model of the sensor of the air handling unit, and outputting to obtain the fault diagnosis result of the sensor in the air handling unit to be predicted;
the method comprises the following steps of constructing a pre-constructed micro fault diagnosis model of the air handling unit sensor, wherein the pre-constructed micro fault diagnosis model of the air handling unit sensor comprises the following specific steps:
optimizing the kernel parameters of the kernel principal component correlation analysis algorithm by using a particle swarm optimization algorithm to obtain an optimized kernel principal component correlation analysis algorithm; replacing Euclidean distances in a kernel principal component analysis algorithm by using correlation distances to optimize Gaussian radial basis kernel functions in the kernel principal component analysis algorithm to obtain the kernel principal component correlation analysis algorithm;
training the optimized kernel principal component correlation analysis algorithm by using sensor historical data of a normal air handling unit to obtain a pre-constructed micro fault diagnosis model of the air handling unit sensor; and the historical data of the sensor of the normal air handling unit comprises normal data and preset fault data.
2. The method according to claim 1, wherein the normal data includes historical data of sensors in a normal air handling unit; the sensor historical data in the normal air handling unit comprises the opening degree of a chilled water valve, the temperature of fresh air, the humidity of the fresh air, the temperature of supplied air, the humidity of supplied air, the temperature of returned air and the humidity of returned air.
3. The method for diagnosing the faults of the sensors in the air handling unit according to claim 1, wherein the preset fault data are acquired after a fault form is preset on the sensors in a normal air handling unit; wherein the preset fault forms comprise 5% -20% of drift fault forms or 5% -20% of deviation fault forms.
4. The method for diagnosing the fault of the sensor in the air handling unit according to claim 1, wherein the real-time data of the sensor in the air handling unit to be predicted is used as the input of a pre-constructed micro fault diagnosis model of the sensor of the air handling unit, and the process of obtaining the fault diagnosis result of the sensor in the air handling unit to be predicted is output, and specifically, the process comprises the following steps:
constructing and obtaining an initial data matrix X according to the real-time data of the sensor in the air handling unit to be predicted N×m ；
Combining the principle that the information entropy is unchanged before and after linear change, and aligning the initial data matrix X N×m Linear transformation is carried out to obtain a homogeneous data matrix Z N×m ；
Respectively carrying out correlation analysis on the initial data matrix X by using the optimized kernel principal component correlation analysis algorithm N×m And the homogeneous data matrix Z N×m Performing dimensionality reduction to obtain an initial data dimensionality reduction result Y N×m And homogenous data dimensionality reduction result Y' N×m ；
Calculating the dimensionality reduction result Y of the initial data N×m And homogenous data dimensionality reduction result Y' N×m Relative distance D between i ；
Correlating the distance D i With a predetermined minimum correlation distance D min Comparing if the related distance D i Less than a predetermined minimum correlation distance D min Then, the principal component contribution ratio pair is utilizedAnd extracting fault characteristics, and outputting to obtain a fault diagnosis result of the sensor in the air handling unit to be predicted.
5. The method of claim 4, wherein the initial data matrix X is a matrix of data values N×m Comprises the following steps:
wherein N is the number of the real-time data of the sensors, and m is the number of the sensors; x' Nm Real-time data of an Nth sensor of the mth sensor;
the homogeneous data matrix Z N×m Comprises the following steps:
wherein, Z ij ' is the result of linear transformation of the ith sensor real-time data of the jth sensor; x is the number of i ′ j Real-time data of an ith sensor of the jth sensor; x is the number of i ′ (j+1) The ith sensor real-time data of the (j + 1) th sensor; x is the number of i ′ m Real-time data of an ith sensor of the mth sensor; x is the number of i ′ 1 Real-time data of the ith sensor of the 1 st sensor.
6. The method of claim 4, wherein the correlation distance D is the distance between the sensor and the air handling unit i Greater than or equal to a preset minimum correlation distance D min Then, the kernel of the kernel principal component correlation analysis algorithm is analyzed by utilizing the particle swarm algorithmUpdating the parameters to obtain an updated kernel principal component correlation analysis algorithm;
respectively carrying out correlation analysis on the initial data matrix X by using the updated kernel principal component correlation analysis algorithm N×m And the homogeneous data matrix Z N×m And carrying out dimension reduction processing and calculating the related distance again.
7. The method for diagnosing the faults of the sensors in the air handling unit according to claim 4, wherein when the fault features are extracted by using the contribution rate of the principal element, the contribution rate CPV (i) of the ith principal element is as follows:
wherein λ is i The characteristic value of the ith principal element, namely the ith kernel function, of the principal component analysis method is obtained; n is the total number of principal elements of the principal component analysis method;
the cumulative contribution CPV of the first p principal elements is:
8. an in-air handling unit sensor fault diagnostic system, comprising:
the data acquisition module is used for acquiring real-time data of a sensor in the air handling unit to be predicted;
the diagnosis output module is used for taking the real-time data of the sensor in the air handling unit to be predicted as the input of a pre-constructed micro fault diagnosis model of the sensor of the air handling unit and outputting to obtain the fault diagnosis result of the sensor in the air handling unit to be predicted;
the method comprises the following steps of constructing a pre-constructed micro fault diagnosis model of the air handling unit sensor, wherein the pre-constructed micro fault diagnosis model of the air handling unit sensor comprises the following steps:
optimizing the kernel parameters of the kernel principal component correlation analysis algorithm by using a particle swarm optimization algorithm to obtain an optimized kernel principal component correlation analysis algorithm; replacing Euclidean distances in a kernel principal component analysis algorithm by using correlation distances to optimize Gaussian radial basis kernel functions in the kernel principal component analysis algorithm to obtain the kernel principal component correlation analysis algorithm;
training the optimized nuclear principal component correlation analysis algorithm by using sensor historical data of a normal air handling unit to obtain a pre-constructed air handling unit sensor tiny fault diagnosis model; and the historical data of the sensor of the normal air handling unit comprises normal data and preset fault data.
9. An in-air handling unit sensor fault diagnostic apparatus, comprising:
a memory for storing a computer program;
a processor for implementing the steps of the method for diagnosing sensor faults in an air handling unit according to any one of claims 1 to 7 when executing said computer program.
10. A computer-readable storage medium, in which a computer program is stored which, when being executed by a processor, carries out the steps of the method for diagnosing faults in sensors of an air handling unit according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211340653.5A CN115526274A (en) | 2022-10-28 | 2022-10-28 | Method, system, equipment and medium for diagnosing faults of sensors in air handling unit |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211340653.5A CN115526274A (en) | 2022-10-28 | 2022-10-28 | Method, system, equipment and medium for diagnosing faults of sensors in air handling unit |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115526274A true CN115526274A (en) | 2022-12-27 |
Family
ID=84702970
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211340653.5A Pending CN115526274A (en) | 2022-10-28 | 2022-10-28 | Method, system, equipment and medium for diagnosing faults of sensors in air handling unit |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115526274A (en) |
-
2022
- 2022-10-28 CN CN202211340653.5A patent/CN115526274A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Zhang et al. | Fault detection and diagnosis of chemical process using enhanced KECA | |
| CN109240274B (en) | Industrial process fault diagnosis method based on high-order correlation | |
| CN111580506A (en) | Industrial process fault diagnosis method based on information fusion | |
| CN113409167B (en) | Water quality abnormity analysis method and device | |
| CN109298633A (en) | Chemical production process fault monitoring method based on adaptive piecemeal Non-negative Matrix Factorization | |
| CN112507479B (en) | Oil drilling machine health state assessment method based on manifold learning and softmax | |
| Chen et al. | A whole building fault detection using weather based pattern matching and feature based PCA method | |
| CN111122811A (en) | Sewage treatment process fault monitoring method of OICA and RNN fusion model | |
| CN115827411A (en) | Online monitoring and operation and maintenance evaluation system and method for automation equipment | |
| CN116383727A (en) | Method, system, equipment and medium for identifying coarse errors in power plant system measurement | |
| CN115060312A (en) | Building material safety monitoring system based on artificial intelligence | |
| CN116627116B (en) | Process industry fault positioning method and system and electronic equipment | |
| CN113033910A (en) | Photovoltaic power generation power prediction method, storage medium and terminal equipment | |
| CN117152152A (en) | Production management system and method for detection kit | |
| CN115526274A (en) | Method, system, equipment and medium for diagnosing faults of sensors in air handling unit | |
| TWI427487B (en) | Method for sampling workpiece for inspection and computer program product performing the same | |
| CN111311025A (en) | Load prediction method based on meteorological similar days | |
| CN116304955A (en) | Switch equipment fault detection method and device, terminal equipment and storage medium | |
| CN115526215A (en) | Method and system for diagnosing, maintaining, analyzing and optimizing rubber pump fault | |
| CN115659271A (en) | Sensor abnormality detection method, model training method, system, device, and medium | |
| CN109190717A (en) | A kind of Multiple Source Sensor fault detection method based on ICA and kNN | |
| CN113188235A (en) | Multi-source fault diagnosis method, system and equipment for sensor in air conditioning system | |
| CN113989632A (en) | Bridge detection method and device for remote sensing image, electronic equipment and storage medium | |
| CN114004138A (en) | Building monitoring method and system based on big data artificial intelligence and storage medium | |
| CN116150666B (en) | Energy storage system fault detection method and device and intelligent terminal |
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 |