Fault diagnosis method and system for floor radiation heating system, medium and equipment

Abstract

The invention discloses a fault diagnosis method and system for a floor radiation heating system, a medium and equipment, and relates to the technical field of fault diagnosis of heating systems, and the method comprises the steps: obtaining a multi-source signal data set of the floor radiation heating system; the method comprises the following steps: performing feature extraction on a multi-source signal by adopting wavelet transform and a time-distributed three-dimensional convolution layer, performing three-dimensional convolution on the extracted features, converting the features into one-dimensional vectors, and performing global average pooling on each signal feature through an average pooling layer to obtain a global statistical feature of each signal channel; activating the compressed feature vectors through a full connection layer to generate a weight vector with the same number as the channels, and multiplying the weight vector by the features extracted from the multi-source signal to realize weight distribution; feature extraction is further carried out on the features after weight adjustment through a two-dimensional convolution kernel, and fault diagnosis is carried out according to an output result; according to the method, system fault diagnosis can be realized in a low-cost and high-accuracy manner.

Description

Technical Field

[0001] The invention relates to the technical field of heating system fault diagnosis, and in particular to a fault diagnosis method, system, medium and equipment for a floor radiation heating system. Background Art

[0002] The floor radiant heating system uses radiant panels as a heating method with radiant heat as the main heat transfer method. It has the advantages of high efficiency and energy saving, good thermal stability, strong comfort, and flexible selection of heat medium. It is generally recognized by the public at home and abroad as a relatively ideal heating method. The floor radiant heating system is prone to various faults such as pipeline leakage, pipeline blockage, gas accumulation in the manifold, and exhaust valve leakage. The type and location of the fault are difficult to detect in a timely and accurate manner. At the least, the heating effect is not ideal and energy is wasted. At the worst, it leads to loss of personnel and property, and even safety problems. Therefore, it is of great significance to carry out research on fault diagnosis methods for floor radiant heating systems.

[0003] At present, a heating equipment fault data processing method based on the Internet of Things has been proposed for the fault diagnosis of the heating pipeline network. Sensors are installed in the heating system to monitor the equipment operation status in real time. The Internet of Things gateway technology is used to upload the data to the cloud platform and perform edge delay filtering and time-frequency transformation fluctuation analysis of state parameters. The analysis results are used to achieve quantitative calculation of the severity of signal faults and classification processing of equipment faults.

[0004] However, the above method requires strong computing power when performing time-frequency transform fluctuation analysis, and has high requirements on the accuracy of the analysis algorithm; when a minor fault occurs in the heating system or the operating environment changes, the error in the fluctuation analysis results may lead to misjudgment of the fault; the edge delay filtering and fluctuation analysis algorithm used may also be affected by factors such as network delay, data noise or sensor accuracy, resulting in inaccurate fault diagnosis results. Summary of the invention

[0005] In view of the deficiency of the prior art that when a minor fault occurs in the heating system or the operating environment changes, the error in the fluctuation analysis result may lead to the misdiagnosis of the fault, the present invention proposes a fault diagnosis method, system, medium and equipment for a floor radiant heating system, which solves the signal energy at different decomposition scales through wavelet transform to form a feature vector, and designs a multi-source signal feature weight adjustment module to realize adaptive weighted fusion of multi-source signal features, and utilizes the feature fusion and fault output module to realize further fusion and feature extraction of the weighted features, enhance the sensitivity of fault features, and improve the accuracy of fault diagnosis, thereby solving the problems existing in the prior art.

[0006] A method for diagnosing a fault in a floor radiant heating system comprises the following steps:

[0007] Collect multi-source signal data of the floor radiant heating system; perform wavelet transform on the multi-source signal data, and use the transformed wavelet coefficients as the characteristics of each signal;

[0008] A time-distributed three-dimensional convolution layer is used to further extract features from the wavelet coefficients, and each extracted signal feature is converted into a one-dimensional vector after three-dimensional convolution. A global average pooling layer is used to perform global average pooling on each one-dimensional vector to obtain the global statistics of each signal channel. The global statistics of each signal channel are activated through a fully connected layer, and multiple weight vectors with the same number of signal channels are generated. Each weight vector is multiplied by the features of each corresponding signal to achieve weight distribution. The signal features after weight distribution are further extracted through a two-dimensional convolution kernel, and the extracted features are classified and output through a fully connected layer and a Softmax layer to obtain the probability value of each fault category.

[0009] The faults of the floor radiant heating system are diagnosed according to the probability value of each category.

[0010] Furthermore, the multi-source signal data set is preprocessed before the wavelet transform is performed on the multi-source signals; the preprocessing process includes data cleaning, data denoising, and data conversion; wherein the multi-source signal data set includes a water pump vibration signal, a water pump noise signal, a hot water pressure pulsation signal, a hot water temperature signal, and a hot water flow signal.

[0011] Furthermore, the multi-source signal data is subjected to wavelet transformation, and the transformation process is expressed as:

[0012]

[0013]

[0014] in, a is the scale parameter; b is the translation parameter; t is the signal on the time axis; is the translation factor; is the wavelet basis function; are the wavelet transform coefficients; is the original signal; is the complex conjugate of the wavelet basis function.

[0015] Furthermore, the three-dimensional convolution layer uses a three-dimensional convolution kernel with a dimension of 3×5×10, the number of convolution kernels is 64, and the activation function used is the ReLU function.

[0016] Furthermore, the weight allocation process is expressed as:

[0017]

[0018] In the formula, For the l ( l =1,2,… n ) signal features extracted from the input signals; Features The weight of is element-wise multiplication; for n Signal characteristics after weight distribution of input signals; For the l ( l =1,2,… n ) signal features are arranged once to form a fault feature vector; n is the number of input signals.

[0019] The present invention also includes a fault diagnosis system for a floor radiant heating system, comprising:

[0020] An acquisition module, used for acquiring multi-source signal data of a floor radiant heating system;

[0021] A feature extraction module, for extracting features from multi-source signal data using wavelet transform and time-distributed three-dimensional convolutional layer;

[0022] The information dimension reduction module is used to remove the redundant information of each one-dimensional vector using a global average pooling layer;

[0023] The weight distribution module is used to activate the global statistics using the fully connected layer, and generate multiple weight vectors corresponding to the number of signal channels to achieve weight distribution;

[0024] The fault output module is used to perform fault classification output using the fully connected layer and the Softmax layer.

[0025] The present invention also includes a fault diagnosis computer device for a floor radiant heating system, comprising: a memory, a processor, and a computer program stored in the memory, wherein the processor implements the steps of the fault diagnosis method for the floor radiant heating system when executing the computer program.

[0026] The present invention also includes a fault diagnosis computer device for a floor radiant heating system, comprising: a memory, a processor, and a computer program stored in the memory, wherein the processor implements the steps of the fault diagnosis method for the floor radiant heating system when executing the computer program.

[0027] The present invention provides a fault diagnosis method, system, medium and equipment for a floor radiant heating system, which has the following beneficial effects:

[0028] The present invention utilizes an average pooling layer to remove redundant data, adjusts weights for multi-source signal features, realizes adaptive weighted fusion of multi-source signal features, and utilizes feature fusion and fault output modules to further fuse and extract weighted features, thereby reducing the amount of data while enhancing the sensitivity of fault features, thereby improving the accuracy of fault diagnosis when a minor fault occurs in the heating system or when the operating environment changes. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 This is a flow chart of a method for diagnosing a fault of a floor radiant heating system according to an embodiment of the present invention;

[0030] Figure 2 4 is a flowchart of adjusting the weights of multi-source signal features in an embodiment of the present invention. DETAILED DESCRIPTION

[0031] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0032] The present invention proposes a fault diagnosis method for floor radiant heating system based on multi-source signal fusion. The signal energy at different decomposition scales is solved by wavelet transform to form a feature vector, and a multi-source signal feature weight adjustment module is designed to realize adaptive weighted fusion of multi-source signal features. The feature fusion and fault output module are used to further fuse and extract the weighted features, enhance the sensitivity of fault features, and improve the accuracy of fault diagnosis. The system fault diagnosis is realized in a low-cost and high-accuracy manner. Figure 1 As shown, the method specifically comprises the following steps:

[0033] S1. A simulation model of floor radiant heating system with hydraulic and thermal coupling is established based on TRNSYS software.

[0034] S1.1. Obtain the basic parameters of the actual building, including building location, function, height, area, structure, material, etc.

[0035] S1.2. Use Sketch Up software to build a three-dimensional model of the building based on the basic parameters of the building.

[0036] S1.3. When dividing the thermal zones of a building, heat users and stairwells are treated as separate thermal zones, and the boundary conditions of all thermal zones are matched.

[0037] S1.4. Set the thermal parameters of the building, input the actual building envelope parameters into the simulation model, and set the active layer parameters of the floor as variables to facilitate model calibration and simulation condition adjustment.

[0038] S1.5. Set ventilation, heating and cooling, heat gain and loss, and initial temperature in the hot zone setting interface. Set the number of cold air infiltration ventilation times as a variable to facilitate model calibration and simulation condition adjustment.

[0039] S1.6. Set floor radiant heating parameters such as water supply temperature, water supply flow, number of hot zone circuits, coil outer diameter, coil wall thickness, coil heat transfer coefficient, etc. Define the water supply temperature and water supply flow as variables and set them as needed.

[0040] S1.7. Establish a system simulation model, including a meteorological parameter module, a geothermal module, a time control module, a data reader module, a display module, a calculator module, a hydraulic condition calculation module, etc.

[0041] S2. Based on the established floor radiant heating system simulation model, various types of common fault simulations are performed.

[0042] S2.1. Based on the established simulation model, the floor radiant heating system is simulated, and fault simulation experiments such as pipe blockage, pipe leakage, exhaust valve leakage, and pipe gas accumulation are carried out to obtain the normal operation and fault operation data of the floor radiant heating system at a low cost.

[0043] S2.2. Use MySQL database to establish a system simulation database. The data tables in the database include basic data tables and common data tables. The basic data tables are used to describe the basic parameter information of the floor radiant heating system; the common data tables are used to record the operating data of the floor radiant heating system, including water supply temperature, pipe flow, pipe pressure, return water temperature, etc.

[0044] S2.3. The normal operation data and fault operation data are transmitted to the MySQL database for storage.

[0045] S3. Use wavelet transform to solve the signal energy at different decomposition scales and arrange them in sequence to form a feature vector.

[0046] S3.1. Preprocess the water pump vibration signal, water pump noise signal, hot water pressure pulsation signal, hot water temperature signal and hot water flow signal, including data cleaning, data denoising and data conversion.

[0047] S3.2. Select a suitable wavelet function by considering factors such as the frequency characteristics, time domain resolution and noise characteristics of the signal, and select a suitable scale parameter based on the characteristic scale of interest in the signal.

[0048] S3.3. Perform wavelet transform on the original signal. By calculating the inner product of the signal and the wavelet function, the wavelet coefficients at different scales are obtained; the calculation expression is as follows:

[0049] (1)

[0050] (2)

[0051] In the formula, a is the scale parameter; b is the translation parameter; t is the signal on the time axis; is the translation factor; is the wavelet basis function; are the wavelet transform coefficients; is the original signal; is the complex conjugate of the wavelet basis function.

[0052] S3.4. Extract wavelet coefficients as features and select some or all of the wavelet coefficients according to requirements. ( For the j Scale and location i The wavelet coefficients on ) are used as the characteristics of each signal.

[0053] S3.5. Further processing is performed on the extracted wavelet coefficients, including dimensionality reduction and normalization.

[0054] S4. Design a multi-source signal feature weight adjustment module.

[0055] S4.1. Use time-distributed 3D convolution (TD-Conv3D) to further extract features of the wavelet coefficients in turn. After three-dimensional convolution, all signal features obtain one-dimensional vectors with the same number as the input signal. The three-dimensional convolution layer uses a three-dimensional convolution kernel with a dimension of 3×5×10. The number of convolution kernels in the three-dimensional convolution layer is 64, and the activation function of the three-dimensional convolution layer uses the Relu function.

[0056] S4.2, the one-dimensional vector passes through the pooling layer with mean pooling as the pooling method, and the feature map of each signal is globally averaged and pooled to obtain the global statistics of each signal channel. The global statistics compressed by the pooling layer are activated through the fully connected layer 1 and the fully connected layer 2 to generate a weight vector with the same number of channels. The weight vector represents the importance of each signal channel, and the distribution of the weight is adjusted by the activation function Sigmoid. The compression and excitation operation generates a one-dimensional weight vector, which is combined with the feature map extracted from the multi-source signal. Multiply them to achieve weight distribution. The number of nodes in the fully connected layer 1 is 64, and the activation function is the Relu function. The number of nodes in the fully connected layer 2 is 1, and the activation function is the sigmoid function, such as Figure 2 The specific mathematical expression of weight distribution is:

[0057] (3)

[0058] In the formula, For the l ( l =1,2,…n) information features extracted from input signals; Features The weight of is element-wise multiplication; The information features after weight assignment for n input signals; For the l ( l =1,2,…n) information features are arranged once to form a fault feature vector; n is the number of input signals.

[0059] S5. Design feature fusion and fault output module.

[0060] S5.1. Further feature extraction is performed on the weighted features by applying a 2D convolution kernel of size 3×3.

[0061] S5.2. Introduce batch normalization (BN), pooling and dropout layers.

[0062] S5.3, the size is ( c , h , w ) is flattened into a one-dimensional vector through the Flatten operation, and then classified and output through the Fully Connected (FC) and Softmax layers to obtain the probability value of each fault category. The fault type with the largest probability value is the final prediction result y.

[0063] The present invention proposes a method for diagnosing faults of a floor radiant heating system based on multi-source signal fusion, which can detect faults existing in the heating system in a low-cost manner and obtain effective and accurate diagnosis results.

[0064] Based on the same inventive concept, the present invention also proposes a fault diagnosis system for a floor radiant heating system, comprising:

[0065] An acquisition module, used for acquiring multi-source signal data of a floor radiant heating system;

[0066] A feature extraction module, for extracting features from multi-source signal data using wavelet transform and time-distributed three-dimensional convolutional layer;

[0067] The information dimension reduction module is used to remove the redundant information of each one-dimensional vector using a global average pooling layer;

[0068] The weight distribution module is used to activate the global statistics using the fully connected layer, and generate multiple weight vectors corresponding to the number of signal channels to achieve weight distribution;

[0069] The fault output module is used to perform fault classification output using the fully connected layer and the Softmax layer.

[0070] The present invention also provides a fault diagnosis computer device for a floor radiant heating system, comprising: a memory, a processor and a computer program stored in the memory, and the processor implements the steps of a fault diagnosis method for a floor radiant heating system when executing the computer program.

[0071] The present invention further provides a readable storage medium, which stores a computer program. The computer program includes program instructions. When the program instructions are executed by a processor, they are used to execute the steps of a fault diagnosis method for a floor radiant heating system.

[0072] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A fault diagnosis method for a floor radiant heating system, characterized in that: The following steps are involved: Collect multi-source signal data of the floor radiant heating system; perform wavelet transform on the multi-source signal data, and use the transformed wavelet coefficients as the characteristics of each signal; A time-distributed three-dimensional convolution layer is used to further extract features from the wavelet coefficients, and each extracted signal feature is converted into a one-dimensional vector after three-dimensional convolution. A global average pooling layer is used to perform global average pooling on each one-dimensional vector to obtain the global statistics of each signal channel. The global statistics of each signal channel are activated through a fully connected layer, and multiple weight vectors with the same number of signal channels are generated. Each weight vector is multiplied by the features of each corresponding signal to achieve weight distribution. The signal features after weight distribution are further extracted through a two-dimensional convolution kernel, and the extracted features are classified and output through a fully connected layer and a Softmax layer to obtain the probability value of each fault category. The faults of the floor radiant heating system are diagnosed according to the probability value of each category.

2. A fault diagnosis method for a floor radiant heating system according to claim 1, characterized in that: The multi-source signal data set is preprocessed before the wavelet transform is performed on the multi-source signal; the preprocessing process includes data cleaning, data denoising, and data conversion; wherein the multi-source signal data set includes a water pump vibration signal, a water pump noise signal, a hot water pressure pulsation signal, a hot water temperature signal, and a hot water flow signal.

3. A fault diagnosis method for a floor radiant heating system according to claim 1, characterized in that: The multi-source signal data is subjected to wavelet transform, and the transformation process is expressed as follows: ； ； in, a is the scale parameter; b is the translation parameter; t is the signal on the time axis; is the translation factor; is the wavelet basis function; are the wavelet transform coefficients; is the original signal; is the complex conjugate of the wavelet basis function.

4. A fault diagnosis method for a floor radiant heating system according to claim 1, characterized in that: The three-dimensional convolution layer uses a three-dimensional convolution kernel with a dimension of 3×5×10, the number of convolution kernels is 64, and the activation function used is the Relu function.

5. The fault diagnosis method of a floor radiant heating system according to claim 1, characterized in that: The weight distribution process is expressed as: ； In the formula, For the l ( l =1,2,… n ) signal features extracted from the input signals; Features The weight of is element-wise multiplication; for n Signal characteristics after weight distribution of input signals; For the l ( l =1,2,… n ) signal features are arranged once to form a fault feature vector; n is the number of input signals.

6. A fault diagnosis system for a floor radiant heating system, characterized in that: include: An acquisition module, used for acquiring multi-source signal data of a floor radiant heating system; A feature extraction module, for extracting features from multi-source signal data using wavelet transform and time-distributed three-dimensional convolutional layer; The information dimension reduction module is used to remove the redundant information of each one-dimensional vector using a global average pooling layer; The weight distribution module is used to activate the global statistics using the fully connected layer, and generate multiple weight vectors corresponding to the number of signal channels to achieve weight distribution; The fault output module is used to perform fault classification output using the fully connected layer and the Softmax layer.

7. A computer device for fault diagnosis of a floor radiant heating system, characterized in that: include: A memory, a processor and a computer program stored in the memory, wherein the processor implements the steps of the fault diagnosis method for the floor radiant heating system according to any one of claims 1 to 5 when executing the computer program.

8. A readable storage medium, characterized in that: The readable storage medium stores a computer program, and the computer program includes program instructions. When the program instructions are executed by a processor, they are used to execute the steps of the fault diagnosis method for the floor radiant heating system according to any one of claims 1 to 5.

Images