Thermal power generating unit auxiliary equipment state monitoring and evaluating system

Abstract

The invention discloses a thermal power generating unit auxiliary equipment state monitoring and evaluation system, and relates to the technical field of thermal power generating unit monitoring and evaluation. Comprising a thermal power generating unit auxiliary equipment layer, a thermal power generating unit auxiliary equipment data transmission layer, a thermal power generating unit auxiliary equipment feature fusion layer, a DBN auxiliary equipment degradation trend prediction layer and a thermal power generating unit auxiliary equipment state evaluation layer. Compared with the prior art, the method adopts an algorithm combining CNN spatial feature extraction and RNN time sequence feature capture to realize fusion of data spatial features and time sequence features, and adopts a DBN algorithm to perform equipment degradation trend prediction on parameters of a steam turbine auxiliary engine system, thereby not only solving the problem of deep training, but also having strong feature learning ability, and improving the accuracy of the deep training. Dynamic tracking of the operation state of the auxiliary equipment of the thermal power generating unit is realized, and the on-off state of the equipment and health assessment of the operation state of the equipment can be displayed in real time.

Description

Technical Field

[0001] This invention relates to the field of thermal power unit monitoring and evaluation technology, and more specifically to a condition monitoring and evaluation system for auxiliary equipment of thermal power units. Background Technology

[0002] Driven by the "dual carbon" goals, the energy structure is rapidly transforming, with large-scale grid connection of intermittent new energy sources such as wind power and photovoltaics. This has fundamentally changed the role of thermal power, shifting from providing stable base load to undertaking flexible peak-shaving and frequency regulation functions. This role change means that thermal power units need to start and stop more frequently, and adjust loads more and more complex and severe operating conditions, which places higher demands on the reliability and adaptability of auxiliary equipment.

[0003] Meanwhile, with the deepening of power market reforms and intensified competition in electricity prices, reducing power generation costs and improving operational economics have become the lifeline for the survival and development of thermal power plants. Auxiliary pumps such as condensate pumps, closed-loop cooling pumps, and circulating water pumps are major consumers of plant power, and their performance degradation directly increases coal consumption for power generation. Therefore, systematic condition monitoring and evaluation of key auxiliary pumps in thermal power units is an inevitable choice for the power industry to promote the development of traditional thermal power towards greater safety, economy, flexibility, and intelligence under the dual background of energy transition and digital transformation.

[0004] Therefore, it is necessary to propose a condition monitoring and evaluation system for auxiliary equipment of thermal power units to solve the above problems. Summary of the Invention

[0005] The purpose of this invention is to address the problems of increased operating costs and insufficient adaptability caused by the performance degradation of thermal power units through systematic condition monitoring and evaluation.

[0006] To achieve the above objectives, the present invention specifically adopts the following technical solution:

[0007] A condition monitoring and assessment system for auxiliary equipment of thermal power units includes a thermal power unit auxiliary equipment layer, a thermal power unit auxiliary equipment data transmission layer, a thermal power unit auxiliary equipment feature fusion layer, a DBN auxiliary equipment deterioration trend prediction layer, and a thermal power unit auxiliary equipment condition assessment layer connected in sequence.

[0008] The auxiliary equipment layer of the thermal power unit is the turbine auxiliary equipment system, which includes a condensate pump equipment unit, a closed-loop cooling pump equipment unit, and a circulating water pump equipment unit for collecting their respective operating status parameter data.

[0009] The auxiliary equipment data transmission layer of the thermal power unit is used to transmit the data collected by the auxiliary equipment layer of the thermal power unit to the feature fusion layer of the auxiliary equipment of the thermal power unit.

[0010] The auxiliary equipment feature fusion layer of the thermal power unit includes a CNN spatial feature extraction unit, an RNN temporal feature capture unit, and a multi-feature fusion unit, which are used to perform spatial feature extraction, temporal feature capture, and feature fusion on the received data.

[0011] The DBN auxiliary equipment degradation trend prediction layer has a built-in DBN algorithm, which is used to predict the degradation trend of equipment based on the fused features.

[0012] The auxiliary equipment status assessment layer of the thermal power unit includes a condensate pump equipment status assessment unit, a closed-loop cooling pump equipment status assessment unit, and a circulating water pump equipment status assessment unit, respectively, corresponding to the condensate pump equipment unit, the closed-loop cooling pump equipment unit, and the circulating water pump equipment unit, which are used to display monitoring data and realize rapid fault diagnosis.

[0013] Furthermore, the parameter data collected by the water pump equipment unit includes bearing vibration, bearing temperature, shaft displacement, outlet pressure, motor current, inlet pressure, and motor temperature;

[0014] The parameters collected by the closed-loop cooling pump unit include bearing vibration, bearing temperature, outlet pressure, motor power, motor current, pump speed, and cooling water temperature.

[0015] The parameters collected by the circulating water pump unit include bearing vibration, bearing temperature, shaft displacement, outlet pressure, motor winding temperature, pump speed, and motor temperature.

[0016] Furthermore, the CNN spatial feature extraction algorithm is implemented through the following steps:

[0017] Solving for convolutional layer feature data:

[0018] ;

[0019] In the formula, This represents the feature map of the i-th layer of a convolutional neural network. Represents convolution calculation, It is the weight matrix of the i-th convolutional kernel. Let i be the bias vector of the i-th layer. This represents the activation function.

[0020] Pooling layer data dimensionality reduction:

[0021] ;

[0022] In the formula, This represents pooling operations.

[0023] The activation function completes the classification:

[0024] ;

[0025] In the formula, These are the weight coefficients in the fully connected layer. It is the value of the i-th neuron in the previous layer. It is the bias of the fully connected layer.

[0026] Furthermore, the RNN spatial feature extraction algorithm is implemented through the following steps:

[0027] Preparation and assembly: ;

[0028] ;

[0029] in, For activation functions; The input is the output layer; W, U, and V are the inputs to the output layer, respectively. The state that was hidden a moment ago and the current hidden state The corresponding weight matrix;

[0030] Calculate the new hidden state: ;

[0031] in, for The hidden state at time step 1, with tanh as the activation function of the hidden layer of the neuron;

[0032] Calculate the current output: ;

[0033] in, for The output at time step 1 is the normalized exponential function softmax().

[0034] Furthermore, the multi-feature fusion unit is configured to fuse the spatial features extracted by the CNN spatial feature extraction unit with the temporal features captured by the RNN temporal feature capture unit, and to achieve dynamic selection of feature fusion strategy.

[0035] Furthermore, the implementation of the DBN algorithm in the DBN auxiliary equipment degradation trend prediction layer includes unsupervised layer-by-layer greedy pre-training and supervised fine-tuning. The unsupervised layer-by-layer greedy pre-training trains each layer of RBM sequentially. The supervised fine-tuning adds a BP network to the last layer of DBN, uses the output of the last layer of RBM as the input of the BP network for training, and optimizes the entire DBN network through error backpropagation.

[0036] Furthermore, the auxiliary equipment status assessment layer of the thermal power unit can intuitively display the overall view and real-time monitoring data, showing the equipment's vibration, temperature monitoring data, sensor battery strength, and wireless signal strength.

[0037] Furthermore, the condensate pump equipment status assessment unit is configured to dynamically track the operating status of the condensate pump and realize rapid diagnosis of cavitation, bearing damage, impeller wear, inlet filter blockage and motor failure of the condensate pump.

[0038] Furthermore, the closed-loop cooling pump equipment status assessment unit is configured to dynamically track the operating status of the closed-loop cooling pump and realize rapid diagnosis of corrosion, mechanical seal failure, and bearing damage faults of the closed-loop cooling pump.

[0039] Furthermore, the circulating water pump equipment status assessment unit is configured to dynamically track the operating status of the circulating water pump and realize rapid diagnosis of faults such as abrasion, biofouling, and excessive vibration of the circulating water pump.

[0040] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0041] 1. This invention employs an algorithm that combines CNN spatial feature extraction with RNN temporal feature capture. This algorithm can simultaneously and collaboratively extract the most valuable spatial features and temporal evolution patterns from the data, achieving the fusion of spatial and temporal features. It has strong robustness and generalization ability, and can significantly improve the accuracy of early fault identification and classification.

[0042] 2. The DBN algorithm employed in this invention, through unsupervised pre-training and supervised fine-tuning, effectively addresses the problems of poor data quality, strong nonlinearity, and high-dimensional data in complex processes, significantly improving the accuracy of soft measurement modeling. It can automatically learn the inherent, high-level feature representations of data from a large amount of unlabeled normal operating condition data, thereby achieving early, unsupervised warnings. Attached Figure Description

[0043] Figure 1 This is a structural block diagram of the present invention.

[0044] Figure 2 This is a real-time monitoring image of the overall appearance of the present invention.

[0045] Figure 3 This is a basic analysis diagram of the impact energy of the bearing.

[0046] Figure labeling: 1. Auxiliary equipment layer of thermal power unit; 2. Data transmission layer of auxiliary equipment of thermal power unit; 3. Feature fusion layer of auxiliary equipment of thermal power unit; 4. Deterioration trend prediction layer of auxiliary equipment of DBN; 5. Condition assessment layer of auxiliary equipment of thermal power unit; 11. Condensate pump equipment unit; 12. Closed-loop cooling pump equipment unit; 13. Circulating water pump equipment unit; 31. CNN spatial feature extraction unit; 32. RNN temporal feature capture unit; 33. Multivariate feature fusion unit; 51. Condition assessment unit of condensate pump equipment; 52. Condition assessment unit of closed-loop cooling pump equipment; 53. Condition assessment unit of circulating water pump equipment. Detailed Implementation

[0047] The technical solutions of the present invention will be clearly and completely described below with reference to the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0048] Please see Figures 1 to 3 A condition monitoring and assessment system for auxiliary equipment of thermal power units includes an auxiliary equipment layer 1, a data transmission layer 2, a feature fusion layer 3, a DBN auxiliary equipment degradation trend prediction layer 4, and a condition assessment layer 5.

[0049] The auxiliary equipment layer 1 of the thermal power unit is the turbine auxiliary equipment system, one of the three major auxiliary equipment systems of the thermal power unit, including condensate pump equipment unit 11, closed cooling pump equipment unit 12 and circulating water pump equipment unit 13.

[0050] The condensate pump unit 11 is used to collect parameter data related to the operating status of the condensate pump, including bearing vibration, bearing temperature, shaft displacement, outlet pressure, motor current, inlet pressure, and motor temperature.

[0051] The closed-loop cooling pump equipment unit 12 is used to collect parameter data related to the operating status of the closed-loop cooling pump, including bearing vibration, bearing temperature, outlet pressure, motor power, motor current, pump speed and cooling water temperature;

[0052] The circulating water pump equipment unit 13 is used to collect parameter data related to the operating status of the circulating water pump, including bearing vibration, bearing temperature, shaft displacement, outlet pressure, motor winding temperature, pump speed and motor temperature.

[0053] The auxiliary equipment data transmission layer 2 of the thermal power unit is used to transmit data related to the condensate pump equipment unit 11, the closed cooling pump equipment unit 12 and the circulating water pump equipment unit 13 in the auxiliary equipment layer 1 of the thermal power unit to the feature fusion layer 3 of the auxiliary equipment of the thermal power unit, so as to achieve the purpose of data transmission.

[0054] The auxiliary equipment feature fusion layer 3 of the thermal power unit extracts, captures and fuses the data features transmitted by the data transmission layer 2 of the auxiliary equipment of the thermal power unit, including CNN spatial feature extraction unit 31, RNN temporal feature capture unit 32 and multi-feature fusion unit 33.

[0055] The CNN spatial feature extraction unit 31 calculates the corresponding entropy value based on the energy distribution of the signal at different time scales and frequencies, which is used to evaluate the complexity and irregularity of the signal.

[0056] Furthermore, the CNN spatial feature extraction unit 31 is implemented through the following steps:

[0057] 1) Solving for convolutional layer feature data:

[0058] ;

[0059] In the formula, This represents the feature map of the i-th layer of a convolutional neural network. Represents convolution calculation, It is the weight matrix of the i-th convolutional kernel. Let i be the bias vector of the i-th layer. This represents the activation function.

[0060] 2) Pooling layer data dimensionality reduction:

[0061] ;

[0062] In the formula, This represents pooling operations.

[0063] 3) The activation function completes the classification:

[0064] ;

[0065] In the formula, These are the weight coefficients in the fully connected layer. It is the value of the i-th neuron in the previous layer. It is the bias of the fully connected layer.

[0066] The RNN temporal feature capture unit 32 completes the effective capture of temporal feature sequence data through preparation and splicing, calculation of new hidden states, and calculation of the current output steps;

[0067] Furthermore, the RNN temporal feature capture unit 32 is implemented through the following steps:

[0068] 1) Preparation and assembly: ;

[0069] ;

[0070] in, For activation functions; The input is the output layer; W, U, and V are the inputs to the output layer, respectively. The state that was hidden a moment ago and the current hidden state The corresponding weight matrix.

[0071] 2) Calculate the new hidden state:

[0072] ;

[0073] in, for The hidden state at time step 1, where tanh is the activation function of the hidden layer of the neuron.

[0074] 3) Calculate the current output:

[0075] ;

[0076] in, for The output at time step 1 is the normalized exponential function softmax().

[0077] The multi-feature fusion unit 33 fuses the data features extracted by the CNN spatial feature extraction unit 31 and the RNN temporal feature capture unit 32, thereby realizing the dynamic selection of feature fusion strategies.

[0078] DBN Auxiliary Equipment Deterioration Trend Prediction Layer 4 uses the built-in DBN algorithm to predict the deterioration trend of the turbine auxiliary equipment system parameters, and can display the equipment start-up and shutdown status and the health assessment of the equipment operating status in real time.

[0079] Furthermore, the DBN auxiliary equipment degradation trend prediction layer 4 is implemented in the following way:

[0080] 1) Unsupervised layer-by-layer greedy pre-training: Train each layer of RBM sequentially to ensure that the feature vector retains as much information as possible when mapped to different feature spaces;

[0081] 2) Supervised Fine-tuning: A backpropagation (BP) network is added to the last layer of the DBN, and the output of the last RBM layer is used as the input to the BP network to train the classifier in a supervised manner. Subsequently, the error is propagated to each RBM layer through backpropagation to optimize the entire DBN network.

[0082] The condensate pump equipment status assessment unit 51 can dynamically track the operating status of the condensate pump and realize rapid diagnosis of cavitation, bearing damage, impeller wear, inlet filter blockage and motor failure of the condensate pump.

[0083] The closed-loop cooling pump equipment status assessment unit 52 can dynamically track the operating status of the closed-loop cooling pump and realize rapid diagnosis of corrosion, mechanical seal failure, and bearing damage faults of the closed-loop cooling pump.

[0084] The circulating water pump equipment status assessment unit 53 can dynamically track the operating status of the circulating water pump and realize rapid diagnosis of faults such as abrasion, biofouling, and excessive vibration of the circulating water pump.

[0085] The auxiliary equipment status assessment layer 5 of the thermal power unit can intuitively display the overall layout and real-time monitoring data, showing the vibration and temperature monitoring data of the equipment, the sensor battery strength and wireless signal strength, realizing the shift from "over-maintenance" to "precise maintenance", reducing unnecessary downtime and disassembly inspection, including condensate pump equipment status assessment unit 51, closed cooling pump equipment status assessment unit 52 and circulating water pump equipment status assessment unit 53.

[0086] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. The scope of patent protection of the present invention shall be determined by the claims. Similarly, any equivalent structural changes made based on the content of the present invention's specification shall also be included within the scope of protection of the present invention.

Claims

1. A condition monitoring and evaluation system for auxiliary equipment of thermal power units, characterized in that: It includes the auxiliary equipment layer of thermal power units, the data transmission layer of auxiliary equipment of thermal power units, the feature fusion layer of auxiliary equipment of thermal power units, the DBN auxiliary equipment deterioration trend prediction layer, and the condition assessment layer of auxiliary equipment of thermal power units, which are connected in sequence. The auxiliary equipment layer of the thermal power unit is the turbine auxiliary equipment system, which includes a condensate pump equipment unit, a closed-loop cooling pump equipment unit, and a circulating water pump equipment unit for collecting their respective operating status parameter data. The auxiliary equipment data transmission layer of the thermal power unit is used to transmit the data collected by the auxiliary equipment layer of the thermal power unit to the feature fusion layer of the auxiliary equipment of the thermal power unit. The auxiliary equipment feature fusion layer of the thermal power unit includes a CNN spatial feature extraction unit, an RNN temporal feature capture unit, and a multi-feature fusion unit, which are used to perform spatial feature extraction, temporal feature capture, and feature fusion on the received data. The DBN auxiliary equipment degradation trend prediction layer has a built-in DBN algorithm, which is used to predict the degradation trend of equipment based on the fused features. The auxiliary equipment status assessment layer of the thermal power unit includes a condensate pump equipment status assessment unit, a closed-loop cooling pump equipment status assessment unit, and a circulating water pump equipment status assessment unit, respectively, corresponding to the condensate pump equipment unit, the closed-loop cooling pump equipment unit, and the circulating water pump equipment unit, which are used to display monitoring data and realize rapid fault diagnosis.

2. The condition monitoring and evaluation system for auxiliary equipment of thermal power units according to claim 1, characterized in that: The parameters collected by the pump unit include bearing vibration, bearing temperature, shaft displacement, outlet pressure, motor current, inlet pressure, and motor temperature. The parameters collected by the closed-loop cooling pump unit include bearing vibration, bearing temperature, outlet pressure, motor power, motor current, pump speed, and cooling water temperature. The parameters collected by the circulating water pump unit include bearing vibration, bearing temperature, shaft displacement, outlet pressure, motor winding temperature, pump speed, and motor temperature.

3. The condition monitoring and evaluation system for auxiliary equipment of thermal power units according to claim 1, characterized in that: The CNN spatial feature extraction algorithm is implemented through the following steps: Solving for convolutional layer feature data: ； In the formula, This represents the feature map of the i-th layer of a convolutional neural network. Represents convolution calculation, It is the weight matrix of the i-th convolutional kernel. Let i be the bias vector of the i-th layer. Represents the activation function; Pooling layer data dimensionality reduction: ； In the formula, Represents pooling operations; The activation function completes the classification: ； In the formula, These are the weight coefficients in the fully connected layer. It is the value of the i-th neuron in the previous layer. It is the bias of the fully connected layer.

4. The condition monitoring and evaluation system for auxiliary equipment of thermal power units according to claim 1, characterized in that: The RNN spatial feature extraction algorithm is implemented through the following steps: Preparation and assembly: ; ； in, For activation functions; The input is the output layer; W, U, and V are the inputs to the output layer, respectively. The state that was hidden a moment ago and the current hidden state The corresponding weight matrix; Calculate the new hidden state: ; in, for The hidden state at time step 1, with tanh as the activation function of the hidden layer of the neuron; Calculate the current output: ; in, for The output at time step 1 is the normalized exponential function softmax().

5. The condition monitoring and evaluation system for auxiliary equipment of thermal power units according to claim 1, characterized in that: The multi-feature fusion unit is configured to fuse the spatial features extracted by the CNN spatial feature extraction unit with the temporal features captured by the RNN temporal feature capture unit, and to achieve dynamic selection of feature fusion strategy.

6. The condition monitoring and evaluation system for auxiliary equipment of thermal power units according to claim 1, characterized in that: The implementation of the DBN algorithm in the DBN auxiliary equipment degradation trend prediction layer includes unsupervised layer-by-layer greedy pre-training and supervised fine-tuning. The unsupervised layer-by-layer greedy pre-training trains each layer of RBM sequentially. The supervised fine-tuning adds a BP network to the last layer of DBN, uses the output of the last layer of RBM as the input of the BP network for training, and optimizes the entire DBN network through error backpropagation.

7. The condition monitoring and evaluation system for auxiliary equipment of thermal power units according to claim 1, characterized in that: The auxiliary equipment status assessment layer of the thermal power unit can intuitively display the overall picture and real-time monitoring data, showing the equipment's vibration, temperature monitoring data, sensor battery strength, and wireless signal strength.

8. The condition monitoring and evaluation system for auxiliary equipment of thermal power units according to claim 1, characterized in that: The condensate pump equipment status assessment unit is configured to dynamically track the operating status of the condensate pump and to quickly diagnose cavitation, bearing damage, impeller wear, inlet filter blockage, and motor failure.

9. The condition monitoring and evaluation system for auxiliary equipment of thermal power units according to claim 1, characterized in that: The closed-loop cooling pump equipment status assessment unit is configured to dynamically track the operating status of the closed-loop cooling pump and realize rapid diagnosis of corrosion, mechanical seal failure, and bearing damage faults of the closed-loop cooling pump.

10. The condition monitoring and evaluation system for auxiliary equipment of thermal power units according to claim 1, characterized in that: The circulating water pump equipment status assessment unit is configured to dynamically track the operating status of the circulating water pump and realize rapid diagnosis of faults such as abrasion, biofouling, and excessive vibration of the circulating water pump.

Images