Power plant thermal control equipment fault prediction method based on big data
By combining time series analysis and deep learning technology, using piezoelectric vibration sensors and CMOS temperature sensors to collect data, the fault prediction challenges of power plant thermal control equipment under excessive jitter and extreme temperature conditions are solved, and high-accurate fault prediction and real-time monitoring are achieved, reducing maintenance costs.
Patent Information
- Application Number
- CN202411788715.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-06
- Publication Date
- 2025-03-25
AI Technical Summary
Failure prediction faces challenges in power plant thermal control equipment under excessive jitter and extreme temperature conditions, and the existing technology is difficult to effectively deal with these situations.
The fault prediction technology based on time series analysis is adopted, combined with the fault prediction technology based on deep learning, and data is collected through piezoelectric vibration sensors and CMOS temperature sensors, and real-time monitoring and prediction are used using the CUSUM method and LSTM model to realize the identification of fault patterns under abnormal vibration and extreme temperatures of the equipment.
It improves the accuracy of fault prediction, realizes real-time monitoring and early warning of equipment status, reduces maintenance costs and management difficulties, and extends the service life of equipment.
Smart Images
Figure CN119691647A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of equipment failure prediction, and in particular to a method for predicting failure of thermal control equipment in a power plant based on big data. Background Art
[0002] With the advancement of science and technology and the continuous maturity of big data technology, the power industry has accumulated a large amount of equipment operation data, environmental parameters and historical fault records. By deeply mining and analyzing these data, the laws and trends of equipment failures are revealed, and an accurate fault prediction model is established. The application of this method can significantly improve the reliability and stability of power plant thermal control equipment. By real-time monitoring of the operating status of the equipment and combining historical fault data, the type, location and time of possible future faults can be predicted, providing important reference for operation and maintenance personnel, enabling them to take measures in advance to prevent and reduce the occurrence of faults, which not only helps to reduce maintenance costs and time, but also improves the operating efficiency and service life of the equipment. In addition, the fault prediction method based on big data can also optimize the configuration of maintenance resources. By analyzing historical maintenance records, the use and effect of maintenance resources can be found, and maintenance tasks and the purchase of updated equipment can be reasonably allocated. These advantages make the application prospect of the power plant thermal control equipment fault prediction method based on big data very promising in the power industry, and it is expected to become an important means of power plant equipment management and maintenance in the future.
[0003] However, when the degree of jitter of thermal control equipment is too large and the temperature is extreme, the fault prediction of thermal control equipment in power plants faces certain challenges. In order to effectively cope with these challenges, the fault prediction technology based on time series analysis is used to timely discover the abnormal vibration state of the equipment. This technology accurately predicts the future vibration state of the equipment by deeply mining the time dependence and periodicity in the vibration data of the equipment, and issues an early warning in time when the abnormal vibration occurs, thereby effectively avoiding the occurrence of equipment failure. At the same time, for the abnormal temperature state of equipment running under extreme temperature conditions, a fault prediction technology based on deep learning is introduced. This technology uses the powerful feature extraction and pattern recognition capabilities of the deep learning model to deeply analyze the temperature data of the equipment, accurately identify the failure mode of the equipment under extreme temperature, and timely discover the abnormal temperature state of the equipment by real-time monitoring of the temperature data and comparing it with the model prediction results, providing a strong guarantee for the safe operation of the power plant. The combined application of these two technologies not only improves the accuracy of fault prediction, but also realizes real-time monitoring and early warning of equipment status, providing a more scientific and efficient method for the maintenance and management of thermal control equipment in power plants. Summary of the invention
[0004] In view of the deficiencies in the prior art, the present invention provides a method for predicting faults of thermal control equipment in a power plant based on big data, which solves the problems in the above-mentioned background technology.
[0005] To achieve the above objectives, the present invention is implemented through the following technical solutions: A method for predicting failures of thermal control equipment in a power plant based on big data, which consists of the following steps: S1, including a piezoelectric vibration sensor and a CMOS temperature sensor, converts the mechanical energy generated by the vibration of the thermal control equipment of the power plant into electrical energy, collects the amplitude data of the thermal control equipment of the power plant, and obtains the operating temperature data of the thermal control equipment of the power plant by converting the measured transistor threshold voltage analog-to-digital into a digital signal; S2. Use the CUSUM method to monitor the changes in the cumulative sum of abnormal vibration times in real time, discover the abnormal vibration state of the equipment in time, establish an abnormal temperature detection model for thermal control equipment in power plants based on LSTM, generate the predicted value of the operating temperature of thermal control equipment in power plants, and predict the failure of thermal control equipment in power plants; S3. Determine whether the vibration state of the thermal control equipment of the power plant is normal based on whether the cumulative total of abnormal vibration detection times of the thermal control equipment of the power plant reaches 3 times. Set the predicted value of the equipment operating temperature output by the LSTM model to take different protective measures when it exceeds the high temperature threshold and the low temperature threshold.
[0006] Furthermore, the process of collecting equipment amplitude data by using a piezoelectric vibration sensor in S1 includes: When the thermal control equipment of a power plant vibrates, the piezoelectric material inside the piezoelectric vibration sensor is subjected to changes in mechanical stress. This stress change causes the crystal structure of the piezoelectric material to change, thereby generating an electric potential difference. The generated potential difference is amplified and processed by the internal circuit of the piezoelectric vibration sensor, and finally converted into an electrical signal, which is directly related to the vibration amplitude of the equipment.
[0007] Furthermore, the process of obtaining the device operating temperature data by measuring the threshold voltage of the transistor by the CMOS temperature sensor in S1 includes: By measuring the temperature-dependent current source current inside the CMOS temperature sensor, a temperature-related signal is obtained. The current is converted into a certain voltage range through a differential amplifier circuit. The output voltage of the differential amplifier circuit is compared with the reference voltage of the reference voltage generator inside the CMOS temperature sensor to obtain an analog voltage. The analog value is then converted into a digital signal by an analog-to-digital converter, and the digital signal is converted into a temperature value through a digital processing algorithm.
[0008] Furthermore, the process of using the CUSUM method in S2 to timely discover abnormal vibration status of the equipment and predict equipment failure includes: The received amplitude data is cleaned and sorted to remove noise and outliers, and the mean, median and standard deviation of the historical amplitude data are calculated. If the data is evenly distributed, the mean and median are close. The reference value is set to be between the mean ± standard deviation. For each amplitude data point, the difference between it and the reference value is calculated, and these difference values are accumulated to form a new cumulative sum sequence. According to the characteristics of the equipment and operation and maintenance experience, a threshold is set in units of the number of cumulative difference values to determine whether the cumulative sum has reached an abnormal level. The system monitors the changes in the cumulative sum in real time. Once the cumulative sum exceeds the set threshold, the system immediately triggers an audible and visual alarm to notify the operation and maintenance personnel to conduct inspection and maintenance.
[0009] Furthermore, the process of training the input layer of the LSTM-based power plant thermal control equipment operating temperature anomaly monitoring model in S2 includes: The LSTM model is trained using historical equipment operating temperature data and known fault records. The input layer of the LSTM model receives equipment operating temperature data from a CMOS temperature sensor, performs data cleaning, data conversion, and data normalization on the received equipment operating temperature data, extracts features from the preprocessed equipment operating temperature data, and the features include temperature value, temperature change rate, and temperature fluctuation range. The statistics of the selected features are calculated and sliding average processing of the time window is performed.
[0010] Furthermore, the process of training the hidden layer of the LSTM-based power plant thermal control equipment operating temperature anomaly monitoring model in S2 includes: The hidden layer of the LSTM model iteratively processes the input time series data through the gating mechanism of its internal LSTM unit, manages the flow and storage of information, gradually learns and captures the changing rules of the equipment operating temperature and potential abnormal patterns, thereby capturing the long-term dependencies of the time series data. The gating mechanism includes a forget gate, an input gate, and an output gate.
[0011] Furthermore, the process of training the hidden layer gating mechanism of the LSTM-based power plant thermal control equipment operating temperature anomaly monitoring model in S2 includes: The forget gate in the hidden layer of the LSTM model determines the information discarded in the cell state of the previous time step. It calculates a forgetting coefficient between 0 and 1 based on the device operating temperature feature representation at the previous time point and the device operating temperature data at the current time point. Each element corresponds to an element in the cell state. When the forgetting coefficient is 0, it means that the element information is completely discarded. When the forgetting coefficient is 1, it means that the element information is completely retained. The input gate determines the addition of new information in the cell state of the current time step. It calculates an input coefficient between 0 and 1 based on the hidden state of the previous time step, the input of the current time step, and a candidate cell state to control the extent to which new information enters the cell state. At the same time, the input gate also generates a candidate cell state that performs a nonlinear transformation on the input of the current time step, so that the model selectively updates the cell state to reflect the new device operating temperature data. The output gate is responsible for calculating the output value based on the cell state and hidden state of the current time step. It first selects information output to the next time step through the sigmoid layer, and then scales the cell state through the tanh layer and multiplies it with the output of the sigmoid layer to obtain the final output value.
[0012] Furthermore, the process of training the output layer of the LSTM-based power plant thermal control equipment operating temperature anomaly monitoring model in S2 to predict the failure of the power plant thermal control equipment includes: The output layer of the LSTM model generates a predicted value of the operating temperature of the power plant thermal control equipment based on the output results of the hidden layer LSTM units, and sets a threshold to determine whether the predicted value indicates that the equipment operating temperature is abnormal. If the predicted value exceeds or is lower than the threshold, the equipment operating temperature is considered abnormal and corresponding maintenance measures need to be taken. If the predicted value is within the threshold range, the equipment operating temperature is considered normal.
[0013] Furthermore, the process of predicting the fault type of the thermal control equipment of the power plant according to the accumulation and change of abnormal vibration detection of the thermal control equipment of the power plant in S3 includes: During the abnormal vibration monitoring process of the thermal control equipment of the power plant, the threshold is set to the cumulative difference value reaching 3 times. When the cumulative sum exceeds 3 times, the vibration state of the thermal control equipment of the power plant is considered to be abnormal and the sound and light alarm is triggered. When the cumulative sum is 3 times or less, the vibration state of the thermal control equipment of the power plant is considered to be normal.
[0014] Furthermore, the process of predicting the fault type of the thermal control equipment of the power plant according to the LSTM model analysis result of the abnormal operating temperature detection in S3 includes: During the abnormal operating temperature monitoring process of the thermal control equipment in the power plant, the equipment operating temperature data collected in real time is input into the LSTM model to obtain the predicted value of the current equipment operating temperature status. The high temperature threshold is set to 85°C and the low temperature threshold is set to 35°C. When the predicted value of the LSTM model exceeds 85°C, the system issues a high temperature alarm to prompt the staff to take cooling measures. When the predicted value of the LSTM model is lower than 35°C, the system issues a low temperature alarm to prompt the staff to perform necessary preheating operations.
[0015] The present invention provides a method for predicting failures of thermal control equipment in power plants based on big data. It has the following beneficial effects: The above-mentioned power plant thermal control equipment fault prediction method based on big data, by introducing fault prediction technology based on time series analysis and fault prediction technology based on deep learning, proposes an effective solution to the fault prediction problem when the equipment jitter is too large and the temperature is extreme. First, this method can improve the accuracy of fault prediction. The fault prediction technology based on time series analysis deeply mines the time dependence and periodicity in the equipment vibration data, while the fault prediction technology based on deep learning accurately identifies the failure mode of the equipment under extreme temperature. The combination of the two makes the fault prediction more accurate and reliable. Secondly, this method realizes real-time monitoring and early warning of the equipment status. By real-time monitoring of equipment vibration and operating temperature data and comparing them with the model prediction results, the abnormal state of the equipment can be discovered in time, providing a strong guarantee for the safe operation of the power plant. Finally, the application of this method also reduces the maintenance cost and management difficulty of thermal control equipment in power plants. By predicting equipment failures in advance, power plants can reasonably arrange maintenance plans to avoid production interruptions and losses caused by sudden equipment failures, while also improving the use efficiency and life of equipment. BRIEF DESCRIPTION OF THE DRAWINGS
[0016] Figure 1 This is a flow chart of a method for predicting failure of thermal control equipment in a power plant based on big data according to an embodiment of the present application. DETAILED DESCRIPTION
[0017] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0018] like Figure 1 As shown, the present invention provides a technical solution: a method for predicting failure of thermal control equipment in a power plant based on big data, which consists of the following steps: S1, including a piezoelectric vibration sensor and a CMOS temperature sensor, converts the mechanical energy generated by the vibration of the thermal control equipment of the power plant into electrical energy, collects the amplitude data of the thermal control equipment of the power plant, and obtains the operating temperature data of the thermal control equipment of the power plant by converting the measured transistor threshold voltage analog-to-digital into a digital signal; S2. Use the CUSUM method to monitor the changes in the cumulative sum of abnormal vibration times in real time, discover the abnormal vibration state of the equipment in time, establish an abnormal temperature detection model for thermal control equipment in power plants based on LSTM, generate the predicted value of the operating temperature of thermal control equipment in power plants, and predict the failure of thermal control equipment in power plants; S3. Determine whether the vibration state of the thermal control equipment of the power plant is normal based on whether the cumulative total of abnormal vibration detection times of the thermal control equipment of the power plant reaches 3 times. Set the predicted value of the equipment operating temperature output by the LSTM model to take different protective measures when it exceeds the high temperature threshold and the low temperature threshold.
[0019] The process of collecting equipment amplitude data by using a piezoelectric vibration sensor in S1 includes: When the thermal control equipment of a power plant vibrates, the piezoelectric material inside the piezoelectric vibration sensor is subjected to changes in mechanical stress. This stress change causes the crystal structure of the piezoelectric material to change, thereby generating an electric potential difference. The generated potential difference is amplified and processed by the internal circuit of the piezoelectric vibration sensor, and finally converted into an electrical signal, which is directly related to the vibration amplitude of the equipment.
[0020] The process of obtaining the device operating temperature data by measuring the threshold voltage of the transistor by the CMOS temperature sensor in S1 includes: By measuring the temperature-dependent current source current inside the CMOS temperature sensor, a temperature-related signal is obtained. The current is converted into a certain voltage range through a differential amplifier circuit. The output voltage of the differential amplifier circuit is compared with the reference voltage of the reference voltage generator inside the CMOS temperature sensor to obtain an analog voltage. The analog value is then converted into a digital signal by an analog-to-digital converter, and the digital signal is converted into a temperature value through a digital processing algorithm.
[0021] In S2, the CUSUM method is used to timely discover the abnormal vibration state of the equipment. The process of predicting equipment failure includes: The received amplitude data is cleaned and sorted to remove noise and outliers, and the mean, median and standard deviation of the historical amplitude data are calculated. If the data is evenly distributed, the mean and median are close. The reference value is set to be between the mean ± standard deviation. For each amplitude data point, the difference between it and the reference value is calculated, and these difference values are accumulated to form a new cumulative sum sequence. According to the characteristics of the equipment and operation and maintenance experience, a threshold is set in units of the number of cumulative difference values to determine whether the cumulative sum has reached an abnormal level. The system monitors the changes in the cumulative sum in real time. Once the cumulative sum exceeds the set threshold, the system immediately triggers an audible and visual alarm to notify the operation and maintenance personnel to conduct inspection and maintenance.
[0022] The process of training the input layer of the LSTM-based power plant thermal control equipment operating temperature anomaly monitoring model in S2 includes: The LSTM model is trained using historical equipment operating temperature data and known fault records. The input layer of the LSTM model receives equipment operating temperature data from a CMOS temperature sensor, performs data cleaning, data conversion, and data normalization on the received equipment operating temperature data, extracts features from the preprocessed equipment operating temperature data, and the features include temperature value, temperature change rate, and temperature fluctuation range. The statistics of the selected features are calculated and sliding average processing of the time window is performed.
[0023] The process of training the hidden layer of the LSTM-based power plant thermal control equipment operating temperature anomaly monitoring model in S2 includes: The hidden layer of the LSTM model iteratively processes the input time series data through the gating mechanism of its internal LSTM unit, manages the flow and storage of information, gradually learns and captures the changing rules of the equipment operating temperature and potential abnormal patterns, thereby capturing the long-term dependencies of the time series data. The gating mechanism includes a forget gate, an input gate, and an output gate.
[0024] The process of training the hidden layer gating mechanism of the LSTM-based power plant thermal control equipment operating temperature anomaly monitoring model in S2 includes: The forget gate in the hidden layer of the LSTM model determines the information discarded in the cell state of the previous time step. It calculates a forgetting coefficient between 0 and 1 based on the device operating temperature feature representation at the previous time point and the device operating temperature data at the current time point. Each element corresponds to an element in the cell state. When the forgetting coefficient is 0, it means that the element information is completely discarded. When the forgetting coefficient is 1, it means that the element information is completely retained. The input gate determines the addition of new information in the cell state of the current time step. It calculates an input coefficient between 0 and 1 based on the hidden state of the previous time step, the input of the current time step, and a candidate cell state to control the extent to which new information enters the cell state. At the same time, the input gate also generates a candidate cell state that performs a nonlinear transformation on the input of the current time step, so that the model selectively updates the cell state to reflect the new device operating temperature data. The output gate is responsible for calculating the output value based on the cell state and hidden state of the current time step. It first selects information output to the next time step through the sigmoid layer, and then scales the cell state through the tanh layer and multiplies it with the output of the sigmoid layer to obtain the final output value.
[0025] The process of training the output layer of the LSTM-based power plant thermal control equipment operating temperature anomaly monitoring model in S2 and predicting the failure of the power plant thermal control equipment includes: The output layer of the LSTM model generates a predicted value of the operating temperature of the power plant thermal control equipment based on the output results of the hidden layer LSTM units, and sets a threshold to determine whether the predicted value indicates that the equipment operating temperature is abnormal. If the predicted value exceeds or is lower than the threshold, the equipment operating temperature is considered abnormal and corresponding maintenance measures need to be taken. If the predicted value is within the threshold range, the equipment operating temperature is considered normal.
[0026] The process of predicting the fault type of the thermal control equipment of the power plant according to the accumulation and change of abnormal vibration detection of the thermal control equipment of the power plant in S3 includes: During the abnormal vibration monitoring process of the thermal control equipment of the power plant, the threshold is set to the cumulative difference value reaching 3 times. When the cumulative sum exceeds 3 times, the vibration state of the thermal control equipment of the power plant is considered to be abnormal and the sound and light alarm is triggered. When the cumulative sum is 3 times or less, the vibration state of the thermal control equipment of the power plant is considered to be normal.
[0027] The process of predicting the fault type of the thermal control equipment of the power plant according to the LSTM model analysis result of the abnormal operating temperature detection in S3 includes: During the abnormal operating temperature monitoring process of the thermal control equipment in the power plant, the equipment operating temperature data collected in real time is input into the LSTM model to obtain the predicted value of the current equipment operating temperature status. The high temperature threshold is set to 85°C and the low temperature threshold is set to 35°C. When the predicted value of the LSTM model exceeds 85°C, the system issues a high temperature alarm to prompt the staff to take cooling measures. When the predicted value of the LSTM model is lower than 35°C, the system issues a low temperature alarm to prompt the staff to perform necessary preheating operations.
[0028] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions. The sentence "includes an element defined by ... does not exclude the existence of other identical elements in the process, method, article or device including the element".
[0029] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.
Claims
1. A method for predicting failures of thermal control equipment in power plants based on big data, characterized in that: The following steps are involved: S1, including a piezoelectric vibration sensor and a CMOS temperature sensor, converts the mechanical energy generated by the vibration of the thermal control equipment of the power plant into electrical energy, collects the amplitude data of the thermal control equipment of the power plant, and obtains the operating temperature data of the thermal control equipment of the power plant by converting the measured transistor threshold voltage analog-to-digital into a digital signal; S2. Use the CUSUM method to monitor the changes in the cumulative sum of abnormal vibration times in real time, discover the abnormal vibration state of the equipment in time, establish an abnormal temperature detection model for thermal control equipment in power plants based on LSTM, generate the predicted value of the operating temperature of thermal control equipment in power plants, and predict the failure of thermal control equipment in power plants; S3. Determine whether the vibration state of the thermal control equipment of the power plant is normal based on whether the cumulative total of abnormal vibration detection times of the thermal control equipment of the power plant reaches 3 times. Set the predicted value of the equipment operating temperature output by the LSTM model to take different protective measures when it exceeds the high temperature threshold and the low temperature threshold.
2. The method for predicting failure of thermal control equipment in a power plant based on big data according to claim 1 is characterized by: The process of collecting equipment amplitude data by using a piezoelectric vibration sensor in S1 includes: When the thermal control equipment of a power plant vibrates, the piezoelectric material inside the piezoelectric vibration sensor is subjected to changes in mechanical stress. This stress change causes the crystal structure of the piezoelectric material to change, thereby generating an electric potential difference. The generated potential difference is amplified and processed by the internal circuit of the piezoelectric vibration sensor, and finally converted into an electrical signal, which is directly related to the vibration amplitude of the equipment.
3. The method for predicting failure of thermal control equipment in a power plant based on big data according to claim 2 is characterized by: The process of obtaining the device operating temperature data by measuring the threshold voltage of the transistor by the CMOS temperature sensor in S1 includes: By measuring the temperature-dependent current source current inside the CMOS temperature sensor, a temperature-related signal is obtained. The current is converted into a certain voltage range through a differential amplifier circuit. The output voltage of the differential amplifier circuit is compared with the reference voltage of the reference voltage generator inside the CMOS temperature sensor to obtain an analog voltage. The analog value is then converted into a digital signal by an analog-to-digital converter, and the digital signal is converted into a temperature value through a digital processing algorithm.
4. The method for predicting failure of thermal control equipment in a power plant based on big data according to claim 3 is characterized by: In S2, the CUSUM method is used to timely discover the abnormal vibration state of the equipment. The process of predicting equipment failure includes: The received amplitude data is cleaned and sorted to remove noise and outliers, and the mean, median and standard deviation of the historical amplitude data are calculated. If the data is evenly distributed, the mean and median are close. The reference value is set to be between the mean ± standard deviation. For each amplitude data point, the difference between it and the reference value is calculated, and these difference values are accumulated to form a new cumulative sum sequence. According to the characteristics of the equipment and operation and maintenance experience, a threshold is set in units of the number of cumulative difference values to determine whether the cumulative sum has reached an abnormal level. The system monitors the changes in the cumulative sum in real time. Once the cumulative sum exceeds the set threshold, the system immediately triggers an audible and visual alarm to notify the operation and maintenance personnel to conduct inspection and maintenance.
5. The method for predicting failure of thermal control equipment in a power plant based on big data according to claim 4 is characterized by: The process of training the input layer of the LSTM-based power plant thermal control equipment operating temperature anomaly monitoring model in S2 includes: The LSTM model is trained using historical equipment operating temperature data and known fault records. The input layer of the LSTM model receives equipment operating temperature data from a CMOS temperature sensor, performs data cleaning, data conversion, and data normalization on the received equipment operating temperature data, extracts features from the preprocessed equipment operating temperature data, and the features include temperature value, temperature change rate, and temperature fluctuation range. The statistics of the selected features are calculated and sliding average processing of the time window is performed.
6. The method for predicting failure of thermal control equipment in a power plant based on big data according to claim 5 is characterized by: The process of training the hidden layer of the LSTM-based power plant thermal control equipment operating temperature anomaly monitoring model in S2 includes: The hidden layer of the LSTM model iteratively processes the input time series data through the gating mechanism of its internal LSTM unit, manages the flow and storage of information, gradually learns and captures the changing rules of the equipment operating temperature and potential abnormal patterns, thereby capturing the long-term dependencies of the time series data. The gating mechanism includes a forget gate, an input gate, and an output gate.
7. The method for predicting failure of thermal control equipment in a power plant based on big data according to claim 5 is characterized by: The process of training the hidden layer gating mechanism of the LSTM-based power plant thermal control equipment operating temperature anomaly monitoring model in S2 includes: The forget gate in the hidden layer of the LSTM model determines the information discarded in the cell state of the previous time step. It calculates a forgetting coefficient between 0 and 1 based on the device operating temperature feature representation at the previous time point and the device operating temperature data at the current time point. Each element corresponds to an element in the cell state. When the forgetting coefficient is 0, it means that the element information is completely discarded. When the forgetting coefficient is 1, it means that the element information is completely retained. The input gate determines the addition of new information in the cell state of the current time step. It calculates an input coefficient between 0 and 1 based on the hidden state of the previous time step, the input of the current time step, and a candidate cell state to control the extent to which new information enters the cell state. At the same time, the input gate also generates a candidate cell state that performs a nonlinear transformation on the input of the current time step, so that the model selectively updates the cell state to reflect the new device operating temperature data. The output gate is responsible for calculating the output value based on the cell state and hidden state of the current time step. It first selects information output to the next time step through the sigmoid layer, and then scales the cell state through the tanh layer and multiplies it with the output of the sigmoid layer to obtain the final output value.
8. The method for predicting failure of thermal control equipment in a power plant based on big data according to claim 7 is characterized by: The process of training the output layer of the LSTM-based power plant thermal control equipment operating temperature anomaly monitoring model in S2 and predicting the failure of the power plant thermal control equipment includes: The output layer of the LSTM model generates a predicted value of the operating temperature of the power plant thermal control equipment based on the output results of the hidden layer LSTM units, and sets a threshold to determine whether the predicted value indicates that the equipment operating temperature is abnormal. If the predicted value exceeds or is lower than the threshold, the equipment operating temperature is considered abnormal and corresponding maintenance measures need to be taken. If the predicted value is within the threshold range, the equipment operating temperature is considered normal.
9. The method for predicting failure of thermal control equipment in a power plant based on big data according to claim 8 is characterized by: The process of predicting the fault type of the thermal control equipment of the power plant according to the accumulation and change of abnormal vibration detection of the thermal control equipment of the power plant in S3 includes: During the abnormal vibration monitoring process of the thermal control equipment of the power plant, the threshold is set to the cumulative difference value reaching 3 times. When the cumulative sum exceeds 3 times, the vibration state of the thermal control equipment of the power plant is considered to be abnormal and the sound and light alarm is triggered. When the cumulative sum is 3 times or less, the vibration state of the thermal control equipment of the power plant is considered to be normal.
10. A method for predicting failures of thermal control equipment in power plants based on big data according to claim 9, characterized in that: The process of predicting the fault type of the thermal control equipment of the power plant according to the LSTM model analysis result of the abnormal operating temperature detection in S3 includes: During the abnormal operating temperature monitoring process of the thermal control equipment in the power plant, the equipment operating temperature data collected in real time is input into the LSTM model to obtain the predicted value of the current equipment operating temperature status. The high temperature threshold is set to 85°C and the low temperature threshold is set to 35°C. When the predicted value of the LSTM model exceeds 85°C, the system issues a high temperature alarm to prompt the staff to take cooling measures. When the predicted value of the LSTM model is lower than 35°C, the system issues a low temperature alarm to prompt the staff to perform necessary preheating operations.