Electric power DCS system fault monitoring device based on deep learning
By adopting the combination of deep learning and peak detection technology in the power DCS system, the rapid and accurate abnormal identification and fault warning of data in complex environments are achieved, the data quality problems are solved, the stability and security of the system are improved, and the stability and security of the system are ensured. The reliability of power supply.
Patent Information
- Application Number
- CN202411148584.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-21
- Publication Date
- 2024-11-12
AI Technical Summary
In complex environments, the data quality of power DCS systems is uneven, and noise and outliers affect the training effect and prediction accuracy of deep learning models. The data volume is large and diverse, resulting in the identification of abnormal data consumes a lot of resources and time, making it difficult to obtain sufficient labeled data training. Models with strong generalization capabilities cannot accurately identify new fault types or abnormal situations.
The fault monitoring device of power DCS system based on deep learning is adopted, including data acquisition module, abnormality detection module, DCS control module and power management platform. The rapid and accurate abnormality recognition of data is achieved through signal conditioning, peak detection and threshold comparison technology, and combined with The CNN+Bi-LSTM model performs real-time analysis to realize automatic diagnosis and early warning of faults.
It effectively reduces the impact of noise and interference on the system, improves the accuracy and efficiency of abnormal data monitoring, improves the stability and safety of the system, significantly reduces unexpected downtime, and ensures the continuity and reliability of power supply.
Smart Images

Figure CN118938874A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of power DCS system fault monitoring, and in particular to a power DCS system fault monitoring device based on deep learning. Background Art
[0002] With the rapid development of the power industry, the power distributed control system (DCS), as a key component of the modern power production process, undertakes the important task of monitoring, controlling and optimizing the power production process. The DCS system integrates a large number of sensors, actuators and controllers, and realizes real-time data collection, processing and transmission through a complex network architecture, thereby ensuring the efficiency and safety of power production.
[0003] In recent years, with the advent of the big data era and the rapid progress of deep learning technology, the power industry has begun to actively explore the application of data mining technology and deep learning algorithms in power DCS systems, in order to further improve the efficiency and accuracy of data analysis. For example, the Chinese invention patent with application publication number CN 118094199 A discloses a power system monitoring and diagnosis method and system. This technical solution collects the operating parameters of power equipment in real time, and uses the fault prediction model to diagnose potential fault nodes, and intelligently sorts the maintenance sequence by evaluating the fault impact index, thereby optimizing the power system fault response process.
[0004] However, in the actual operation of the power DCS system, due to the complex and changeable power production environment and the large number of widely distributed equipment, the system is easily disturbed by various external and internal factors, resulting in uneven quality of the collected data, noise, missing values, outliers and other problems, which will affect the training effect and prediction accuracy of the deep learning model. In the process of abnormal data monitoring, due to the huge amount of data generated by the power DCS system and the diverse data types, including real-time data, historical data, alarm information, etc., it takes a lot of computing resources and time to accurately identify abnormal data in such a large data set. In addition, deep learning models often rely on a large amount of labeled data during training. In practical applications, it is difficult to obtain enough labeled data to train a model with strong generalization ability, resulting in the model's inability to accurately identify new fault types or abnormal situations.
[0005] In view of the above situation, the present invention aims to provide a new solution to solve this problem. Summary of the invention
[0006] In view of the above situation, in order to overcome the defects of the prior art, the purpose of the present invention is to provide a power DCS system fault monitoring device based on deep learning.
[0007] The technical solution is: a power DCS system fault monitoring device based on deep learning, including:
[0008] A data acquisition module, used to collect system power data through a sensor array, and perform signal conditioning and analog-to-digital conversion processing on the power data;
[0009] An abnormality detection module is used to collect the peak voltage of the detection signal of the sensor array and then perform threshold comparison processing; if the peak voltage is between the high threshold voltage VTHH and the low threshold voltage VTHL, the abnormality detection module outputs a low level signal, indicating that the data detection is normal; if the peak voltage exceeds the threshold range, the abnormality detection module outputs a high level signal, indicating that the data detection is abnormal;
[0010] A DCS control module, used to receive the output data of the data acquisition module and the output signal of the abnormality detection module, and dynamically adjust the working status of the data acquisition module and the abnormality detection module by sending control instructions;
[0011] A communication module, used for data transmission and communication between the DCS control module and the power management platform;
[0012] The power management platform uses deep learning algorithms to perform real-time analysis of received data to achieve automatic diagnosis and early warning of faults.
[0013] Preferably, the anomaly detection module includes:
[0014] A multi-channel data collector, whose input end is connected to the sensor array and is used to output the detection signals of the sensor array in sequence through different channels; whose control end is connected to the DCS control module and selects the channel according to the control instruction of the DCS control module;
[0015] A peak detection circuit is used to extract the peak value of the signal from the sensor signal;
[0016] The threshold comparison circuit is used to compare the peak voltage output by the peak detection circuit with a preset threshold value, and send the comparison result to the DCS control module.
[0017] Preferably, the peak detection circuit includes an op amp AR1 and an op amp AR2, the in-phase input terminal of the op amp AR1 is connected to the signal output terminal of the multi-channel data acquisition device through a resistor R1 and is grounded through a capacitor C1, the inverting input terminal of the op amp AR1 is connected to the anode of a diode VD1 and one end of a resistor R2, the output terminal of the op amp AR1 is connected to the cathode of the diode VD1 and the anode of a diode VD2, the cathode of the diode VD2 is connected to the in-phase input terminal of the op amp AR2 and is grounded through a parallel resistor R3 and capacitor C2, the inverting input terminal and the output terminal of the op amp AR2 are connected to the other end of the resistor R2, the cathode of the voltage zener diode DZ1 and the input terminal of the threshold comparison circuit, and the anode of the voltage zener diode DZ1 is grounded.
[0018] Preferably, the threshold comparison circuit includes an op amp AR3, an op amp AR4 and a photoelectric coupler U1, the in-phase input terminal of the op amp AR3 is connected to the adjustment terminal of the adjustable resistor RP1, and is connected to the power supply terminal of the threshold power supply circuit through a resistor R5; the inverting input terminal of the op amp AR4 is connected to the other end of the adjustable resistor RP1, and is grounded through a resistor R6; the inverting input terminal of the op amp AR3 and the in-phase input terminal of the op amp AR4 are connected to the output terminal of the op amp AR2 through a resistor R4; the output terminals of the op amp AR3 and the op amp AR4 are connected to the anode of the transmitting terminal of the photoelectric coupler U1 through diodes VD3 and VD4 respectively, the cathode of the transmitting terminal of the photoelectric coupler U1 is grounded, the collector of the receiving terminal of the photoelectric coupler U1 is connected to the DCS control module, and is connected to the power supply VCC through a resistor R11, and the emitter of the receiving terminal of the photoelectric coupler U1 is grounded.
[0019] Preferably, the threshold power supply circuit includes a +12V power supply, the +12V power supply is connected to the anode of the voltage zener diode DZ2 and one end of the adjustable resistor RP2 through a resistor R7, the cathode of the voltage zener diode DZ2 and the other end of the adjustable resistor RP2 are grounded, the adjustment end of the adjustable resistor RP2 is connected to the in-phase input end of the operational amplifier AR5 and one end of the resistor R9 through a resistor R8, the inverting input end of the operational amplifier AR5 is grounded through a resistor R1O, and the output end of the operational amplifier AR5 is connected to the other end of the resistor R9 and the power supply end of the threshold power supply circuit.
[0020] Preferably, the data acquisition module further includes:
[0021] A signal conditioning circuit, used for amplifying and harmonic suppression processing of the detection signal of the sensor array;
[0022] The data acquisition card is used to perform analog-to-digital conversion on the signal processed by the signal conditioning circuit and transmit the converted digital signal to the DCS control module; the DCS control module manages the working state of the data acquisition card by sending control instructions.
[0023] Preferably, the signal conditioning circuit comprises:
[0024] A differential amplification unit, used for performing differential amplification processing on the detection signal of the sensor array;
[0025] A harmonic suppression unit, used for suppressing harmonic components in the differentially amplified signal;
[0026] A dynamic compensation unit, used to collect the signal after harmonic suppression, and feed it back to the differential amplifier unit after processing through a compensation adjustment mechanism to eliminate signal fluctuations;
[0027] The differential amplifier unit includes an operational amplifier AR6, an inverting input terminal of the operational amplifier AR6 is connected to the first detection signal input port through a resistor R21 and is grounded through a capacitor C11; a non-inverting input terminal of the operational amplifier AR6 is connected to the second detection signal input port through a resistor R22 and is grounded through a capacitor C12.
[0028] Preferably, the harmonic suppression unit includes an inductor L1, a capacitor C13, a capacitor C15 and a resistor R26, one end of the inductor L1 and the capacitor C13 are connected to the output end of the operational amplifier AR6, the other end of the inductor L1 and the capacitor C13 are connected to one end of the resistor R26 and the data acquisition card, and the other end of the resistor R26 is grounded through the capacitor C15.
[0029] Preferably, the dynamic compensation unit includes an op amp AR7, an inverting input terminal of the op amp AR7 is connected to one end of a resistor R26 via a resistor R25, and an output terminal of the op amp AR7 is connected to a resistor R24 and a capacitor C14 in parallel, a non-inverting input terminal of the op amp AR7 is grounded, and the output terminal of the op amp AR7 is also connected to the non-inverting input terminal of the op amp AR6 via a resistor R23.
[0030] Preferably, the power management platform uses a deep learning model based on CNN+Bi-LSTM to perform real-time analysis of power data to achieve automatic diagnosis and early warning of faults.
[0031] Through the above technical solution, the beneficial effects of the present invention are:
[0032] 1. This application realizes comprehensive and real-time collection of power data through the data acquisition module, and uses the signal conditioning circuit to ensure the collection quality of power data, effectively reducing the impact of noise and interference on the system. The anomaly detection module adopts peak detection and threshold comparison technology to realize fast and accurate anomaly identification of power detection data, significantly reduce system resource consumption, improve operation efficiency and stability, and ensure the accuracy of abnormal data monitoring.
[0033] 2. The power DCS system fault monitoring solution of this application, which combines a deep learning model with an anomaly detection module, can fully utilize the powerful ability of deep learning in processing complex nonlinear relationships and pattern recognition, and at the same time, with the help of the rapid response characteristics of the anomaly detection module, it can realize real-time monitoring and accurate early warning of power system anomalies, thereby significantly improving the stability and safety of the system, effectively preventing potential failures, reducing unexpected downtime, and ensuring the continuity and reliability of power supply. BRIEF DESCRIPTION OF THE DRAWINGS
[0034] Figure 1 It is a structural block diagram of the system module of the present invention.
[0035] Figure 2 It is a structural block diagram of the anomaly detection module in the present invention.
[0036] Figure 3 It is a circuit schematic diagram of the abnormality detection module in the present invention.
[0037] Figure 4 It is a structural block diagram of the data acquisition module in the present invention.
[0038] Figure 5 This is a schematic diagram of the signal conditioning circuit in the present invention.
[0039] Figure 6 It is a flow chart of the data training method of the deep learning model based on CNN+Bi-LSTM of the present invention. DETAILED DESCRIPTION
[0040] The above and other technical contents, features and effects of the present invention are described in detail below with reference to the attached Figure 1 To Attachment Figure 6 The detailed description of the embodiments will clearly show that the structural contents mentioned in the following embodiments are all based on the drawings in the specification.
[0041] Exemplary embodiments of the present invention will be described below with reference to the accompanying drawings.
[0042] like Figure 1 As shown, the power DCS system fault monitoring device based on deep learning includes:
[0043] A data acquisition module is used to collect system power data through a sensor array and perform signal conditioning and analog-to-digital conversion processing on the power data;
[0044] The abnormality detection module is used to collect the peak voltage of the detection signal of the sensor array and then perform threshold comparison processing; if the peak voltage is between the high threshold voltage VTHH and the low threshold voltage VTHL, the abnormality detection module outputs a low-level signal, indicating that the data detection is normal; if the peak voltage exceeds the threshold range, the abnormality detection module outputs a high-level signal, indicating that the data detection is abnormal;
[0045] A DCS control module is used to receive the output data of the data acquisition module and the output signal of the abnormality detection module, and dynamically adjust the working status of the data acquisition module and the abnormality detection module by sending control instructions;
[0046] Communication module, used for data transmission and communication between DCS control module and power management platform;
[0047] The power management platform uses deep learning algorithms to perform real-time analysis of received data to achieve automatic diagnosis and early warning of faults.
[0048] In the above, in order to improve the detection efficiency and accuracy of abnormal data by the power DCS system, such as Figure 2 As shown, the anomaly detection module includes:
[0049] The multi-channel data collector has an input end connected to the sensor array and is used to output the detection signals of the sensor array in sequence through different channels; its control end is connected to the DCS control module and selects the channel according to the control instruction of the DCS control module;
[0050] A peak detection circuit is used to extract the peak value of the signal from the sensor signal;
[0051] The threshold comparison circuit is used to compare the peak voltage output by the peak detection circuit with a preset threshold value, and send the comparison result to the DCS control module.
[0052] Specifically, each sensor in the sensor array converts the various parameters (such as current, voltage, temperature, etc.) detected in the power production process into electrical signals and transmits them to the multi-channel data collector. The multi-channel data collector selects the corresponding channel for data collection according to the control instructions of the DCS control module and sends the collected signals to the peak detection circuit.
[0053] like Figure 3As shown, the peak detection circuit includes an operational amplifier AR1 and an operational amplifier AR2, the non-inverting input terminal of the operational amplifier AR1 is connected to the signal output terminal of the multi-channel data acquisition device through a resistor R1, and is grounded through a capacitor C1, the inverting input terminal of the operational amplifier AR1 is connected to the anode of a diode VD1 and one end of a resistor R2, the output terminal of the operational amplifier AR1 is connected to the cathode of the diode VD1 and the anode of a diode VD2, the cathode of the diode VD2 is connected to the non-inverting input terminal of the operational amplifier AR2, and is grounded through a parallel resistor R3 and a capacitor C2, the inverting input terminal and the output terminal of the operational amplifier AR2 are connected to the other end of the resistor R2, the cathode of a voltage zener diode DZ1 and the input terminal of a threshold comparison circuit, and the anode of the voltage zener diode DZ1 is grounded.
[0054] In the specific working process of the peak detection circuit, when the input signal voltage Vin is greater than the output signal voltage Vo1, AR1 outputs a forward voltage, the diode VD1 is cut off, the diode VD2 is turned on, and the capacitor C2 is quickly charged to Vin; when Vin is less than Vo1, AR1 outputs a negative voltage, the diode VD1 is turned on, the diode VD2 is cut off, and the capacitor C2 is discharged through the resistor R3. Through the above working process, the peak detection circuit can capture and store the peak voltage in the sensor detection signal. When the input signal changes, the circuit can automatically update the stored peak voltage and output it to the threshold comparison circuit, which is very effective in identifying transient or sudden abnormalities in the power production process.
[0055] The threshold comparison circuit includes an operational amplifier AR3, an operational amplifier AR4 and a photoelectric coupler U1, wherein the in-phase input terminal of the operational amplifier AR3 is connected to the adjustment terminal of the adjustable resistor RP1 and is connected to the power supply terminal of the threshold power supply circuit through a resistor R5; the inverting input terminal of the operational amplifier AR4 is connected to the other end of the adjustable resistor RP1 and is grounded through a resistor R6; the inverting input terminal of the operational amplifier AR3 and the in-phase input terminal of the operational amplifier AR4 are connected to the output terminal of the operational amplifier AR2 through a resistor R4; the output terminals of the operational amplifier AR3 and the operational amplifier AR4 are connected to the anode of the transmitting terminal of the photoelectric coupler U1 through a diode VD3 and a diode VD4 respectively, the cathode of the transmitting terminal of the photoelectric coupler U1 is grounded, the collector of the receiving terminal of the photoelectric coupler U1 is connected to the DCS control module, and is connected to the power supply VCC through a resistor R11, and the emitter of the receiving terminal of the photoelectric coupler U1 is grounded.
[0056] During the operation of the threshold comparison circuit, the series connection of resistors R5, R6 and the adjustable resistor RP1 divides the threshold power supply. By adjusting RP1, the voltage division ratio can be changed, thereby forming a high threshold voltage VTHH at the non-inverting input of the op amp AR3 and a low threshold voltage VTHL at the inverting input of the op amp AR4. The op amp AR3 and the op amp AR4 form a window comparator. The peak voltage Vo1 output by the peak detection circuit is sent to the inverting input of the op amp AR3 and the non-inverting input of the op amp AR4 after the amplitude is stabilized by the voltage stabilizing diode DZ1. According to the principle of the window comparator, when the input peak voltage Vo1 is between the two threshold voltages, that is, VTHL≤Vo1≤VTHH, Vo2 outputs a high level; when the input peak voltage Vo1 exceeds the above threshold range, that is, Vo1>VTHH, or Vo1<VTHL, Vo2 outputs a low level.
[0057] In order to prevent external electrical interference from affecting the detection results of the threshold comparison circuit, a photocoupler U1 is set at the output end of the window comparator to electrically isolate the output level. If Vo2 outputs a high level, it means that the data detection is normal at this time, the photocoupler U1 is turned on, and the abnormal detection signal receiving pin (P1.0) of the DCS control module receives a low level signal; if Vo2 outputs a low level, it means that the data detection is abnormal at this time, the photocoupler U1 is turned off, and the abnormal detection signal receiving pin (P1.0) of the DCS control module receives a high level. The DCS control module can quickly determine whether the power detection data is abnormal based on the high and low levels output by the abnormal detection module, which greatly reduces the consumption of system computing resources and time, improves the overall operation efficiency and stability of the system, and effectively ensures the accuracy of abnormal data monitoring.
[0058] In the above, the threshold power supply circuit includes a +12V power supply, the +12V power supply is connected to the anode of the voltage zener diode DZ2 and one end of the adjustable resistor RP2 through a resistor R7, the cathode of the voltage zener diode DZ2 and the other end of the adjustable resistor RP2 are grounded, the adjustment end of the adjustable resistor RP2 is connected to the in-phase input end of the op amp AR5 and one end of the resistor R9 through a resistor R8, the inverting input end of the op amp AR5 is grounded through a resistor R1O, and the output end of the op amp AR5 is connected to the other end of the resistor R9 and the power supply end of the threshold power supply circuit. Among them, the voltage zener diode DZ2 is used to stabilize the voltage and ensure the stability of the power supply of its back-end circuit; the adjustment end of the adjustable resistor RP2 is connected to the in-phase input end of the op amp AR5 through a resistor R8, allowing the user to change the voltage of the in-phase input end of the op amp AR5 by adjusting RP2, thereby adjusting the output voltage of the entire threshold power supply circuit. As a voltage follower, the op amp AR5 outputs its input voltage to the power supply end of the threshold power supply circuit, providing a stable threshold power supply voltage for the subsequent threshold comparison circuit.
[0059] In another embodiment of the present application, when the system performs data collection, the data collection module further includes:
[0060] A signal conditioning circuit, used for amplifying and harmonic suppression processing of the detection signal of the sensor array;
[0061] The data acquisition card is used to perform analog-to-digital conversion on the signal processed by the signal conditioning circuit and transmit the converted digital signal to the DCS control module; the DCS control module manages the working status of the data acquisition card by sending control instructions, including starting / stopping acquisition, setting sampling frequency, selecting acquisition channels, etc.
[0062] Specifically, in order to ensure the accuracy of power data collection, such as Figure 4 As shown, the signal conditioning circuit includes:
[0063] A differential amplification unit, used for differential amplification of detection signals of the sensor array;
[0064] A harmonic suppression unit, used for suppressing harmonic components in the differentially amplified signal;
[0065] The dynamic compensation unit is used to collect the signal after harmonic suppression, and feed it back to the differential amplifier unit after processing through the compensation adjustment mechanism to eliminate signal fluctuations.
[0066] Among them, Figure 5 As shown, the differential amplifier unit includes an operational amplifier AR6, an inverting input terminal of the operational amplifier AR6 is connected to the first detection signal input port through a resistor R21, and is grounded through a capacitor C11; a non-inverting input terminal of the operational amplifier AR6 is connected to the second detection signal input port through a resistor R22, and is grounded through a capacitor C12.
[0067] The harmonic suppression unit includes an inductor L1, a capacitor C13, a capacitor C15 and a resistor R26. One end of the inductor L1 and the capacitor C13 is connected to the output end of the operational amplifier AR6, and the other end of the inductor L1 and the capacitor C13 is connected to one end of the resistor R26 and the data acquisition card. The other end of the resistor R26 is grounded through the capacitor C15.
[0068] The dynamic compensation unit includes an operational amplifier AR7, an inverting input terminal of the operational amplifier AR7 is connected to one end of a resistor R26 via a resistor R25, and an output terminal of the operational amplifier AR7 is connected to a parallel resistor R24 and a capacitor C14, a non-inverting input terminal of the operational amplifier AR7 is grounded, and an output terminal of the operational amplifier AR7 is also connected to a non-inverting input terminal of an operational amplifier AR6 via a resistor R23.
[0069] The specific working process and principle of the signal conditioning circuit are as follows: First, the differential amplifier unit uses the operational amplifier AR6 as the main amplifier to differentially amplify the detection signal of the sensor array, thereby effectively suppressing the common mode noise in the signal acquisition process and improving the signal-to-noise ratio. Secondly, the harmonic suppression unit uses the LC-RC filter network to suppress the harmonic components in the differentially amplified signal to avoid the harmonics generated by various nonlinear loads in the power system causing waveform distortion of the collected signal, thereby improving the signal quality. Finally, the dynamic compensation unit realizes feedback control through the operational amplifier AR7, collects the signal after harmonic suppression, and forms negative feedback on the signal amplification process through the phase compensation network composed of resistor R24 and capacitor C14 to realize compensation adjustment of the differential amplifier unit. When the signal fluctuates, the dynamic compensation unit will adjust its output and eliminate the signal fluctuation by changing the input reference point of the differential amplifier unit. The signal conditioning circuit realizes high-precision acquisition of power data through the coordinated work of the three units of differential amplification, harmonic suppression and dynamic compensation, providing more reliable data support for the subsequent processing and analysis of the system.
[0070] The DCS control module is the core control unit of the system, responsible for receiving the output of the data acquisition module and the anomaly detection module. The DCS control module analyzes and processes the received data to determine whether the system's operating status is normal and whether it is necessary to adjust the working parameters or modes of the data acquisition module and the anomaly detection module according to the current situation. The DCS control module sends the processed data to the power management platform through the communication module, including real-time power parameters, equipment status information, anomaly detection results, etc. In the specific setting, the communication module integrates the communication network port and the 5G module, which can provide both wired and wireless high-speed and stable communication methods, ensuring that the data transmission of the power data acquisition and monitoring system has both the stability and reliability of the wired network and the wide coverage, low latency and flexibility of the 5G network, thereby meeting the complex and changeable power system monitoring needs.
[0071] In another embodiment of the present application, the power management platform uses a combined model based on a convolutional neural network (CNN) and a bidirectional long short-term memory network (Bi-LSTM) to perform real-time analysis of power data to achieve automatic diagnosis and early warning of faults, such as Figure 6 As shown, the specific steps include:
[0072] Data preprocessing: First, the input circuit data is decomposed into multiple intrinsic mode components (IMFs) with limited bandwidth using variational mode decomposition (VMD) technology, which can separate the noise and signal in the data and improve the prediction accuracy. The goal of VMD is to minimize the sum of the frequency bandwidths of each IMFs while ensuring the balance of the modal demodulation of each IMF.
[0073] Feature extraction: For each decomposed modal component, CNN is used to extract the power consumption feature information. CNN can capture local features and reduce the number of network parameters through its convolution layer, which is suitable for processing high-dimensional data; the convolution operation in CNN can be expressed as:
[0074] f(x)=Σ i w i *x i +b
[0075] Among them, x is the input data, w i is the weight of the convolution kernel, and b is the bias term.
[0076] Time series modeling: Bi-LSTM is used to further model the feature information. Due to its memory unit, Bi-LSTM is able to capture long-term dependencies and process sequence data with forward and backward dependencies. The forward and backward propagation of Bi-LSTM can be updated by the following formula:
[0077] h t =tanh(W h *h t-1 +V h *x t +b h
[0078] y t =W y *h t +b y
[0079] Among them, h t is the hidden state at time step t, x t is the input, w h , V h , W y is the weight matrix, b h , b y is the bias term.
[0080] Model training: The CNN-BiLSTM model is trained with a large amount of historical power data. In this stage, the model learns the patterns and trends in the load data in order to make accurate predictions.
[0081] Prediction and reconstruction: For each modal component, the CNN-BiLSTM model provides a prediction result. Then, the prediction results of all modal components are combined to reconstruct the final load forecast value.
[0082] Model evaluation: Use indicators such as mean square error (MSE) or mean absolute error (MAE) to evaluate the performance of the model. At the same time, visualization tools can be used to display the comparison chart between the predicted results and the actual data to intuitively evaluate the accuracy and stability of the model.
[0083] Real-time monitoring and early warning: Deploy the trained model to the power DCS system for real-time data monitoring and fault prediction. Once an abnormal pattern is detected, the system will issue an early warning so that preventive measures can be taken.
[0084] Model updating: Regularly update the model with new data to ensure its accuracy and adaptability.
[0085] In summary, the present application realizes comprehensive and real-time collection of power data through the data acquisition module, and uses the signal conditioning circuit to ensure the collection quality of power data, effectively reducing the impact of noise and interference on the system. The anomaly detection module adopts peak detection and threshold comparison technology to realize fast and accurate anomaly recognition of power detection data, significantly reduce system resource consumption, improve operation efficiency and stability, and ensure the accuracy of abnormal data monitoring. The DCS control module dynamically adjusts the working state of the system according to the results of data acquisition and anomaly detection to realize intelligent monitoring and management. At the same time, the power DCS system fault monitoring solution combining the deep learning model and the anomaly detection module can make full use of the powerful ability of deep learning in processing complex nonlinear relationships and pattern recognition, and at the same time, with the help of the rapid response characteristics of the anomaly detection module, it can realize real-time monitoring and accurate early warning of power system anomalies, thereby significantly improving the stability and safety of the system, effectively preventing potential failures, reducing unexpected downtime, and ensuring the continuity and reliability of power supply.
[0086] The above is a further detailed description of the present invention in combination with a specific implementation method, and it cannot be determined that the specific implementation of the present invention is limited to this; for technical personnel in the technical field to which the present invention belongs and related technical fields, based on the technical solution of the present invention, the expansion and replacement of operating methods and data should all fall within the protection scope of the present invention.
Claims
1. A power DCS system fault monitoring device based on deep learning, characterized in that: include: A data acquisition module, used to collect system power data through a sensor array, and perform signal conditioning and analog-to-digital conversion processing on the power data; An abnormality detection module is used to collect the peak voltage of the detection signal of the sensor array and then perform threshold comparison processing; if the peak voltage is between the high threshold voltage VTHH and the low threshold voltage VTHL, the abnormality detection module outputs a low level signal, indicating that the data detection is normal; if the peak voltage exceeds the threshold range, the abnormality detection module outputs a high level signal, indicating that the data detection is abnormal; A DCS control module, used to receive the output data of the data acquisition module and the output signal of the abnormality detection module, and dynamically adjust the working status of the data acquisition module and the abnormality detection module by sending control instructions; A communication module, used for data transmission and communication between the DCS control module and the power management platform; The power management platform uses deep learning algorithms to perform real-time analysis of received data to achieve automatic diagnosis and early warning of faults.
2. The power DCS system fault monitoring device based on deep learning according to claim 1 is characterized in that: The anomaly detection module comprises: A multi-channel data collector, whose input end is connected to the sensor array and is used to output the detection signals of the sensor array in sequence through different channels; whose control end is connected to the DCS control module and selects the channel according to the control instruction of the DCS control module; A peak detection circuit is used to extract the peak value of the signal from the sensor signal; The threshold comparison circuit is used to compare the peak voltage output by the peak detection circuit with a preset threshold value, and send the comparison result to the DCS control module.
3. The power DCS system fault monitoring device based on deep learning according to claim 2 is characterized in that: The peak detection circuit includes an operational amplifier AR1 and an operational amplifier AR2. The in-phase input terminal of the operational amplifier AR1 is connected to the signal output terminal of the multi-channel data acquisition device through a resistor R1 and is grounded through a capacitor C1. The inverting input terminal of the operational amplifier AR1 is connected to the anode of a diode VD1 and one end of a resistor R2. The output terminal of the operational amplifier AR1 is connected to the cathode of the diode VD1 and the anode of a diode VD2. The cathode of the diode VD2 is connected to the in-phase input terminal of the operational amplifier AR2 and is grounded through a parallel resistor R3 and a capacitor C2. The inverting input terminal and the output terminal of the operational amplifier AR2 are connected to the other end of the resistor R2, the cathode of a voltage stabilizing diode DZ1 and the input terminal of the threshold comparison circuit. The anode of the voltage stabilizing diode DZ1 is grounded.
4. The power DCS system fault monitoring device based on deep learning according to claim 3 is characterized in that: The threshold comparison circuit includes an operational amplifier AR3, an operational amplifier AR4 and a photoelectric coupler U1, wherein the in-phase input terminal of the operational amplifier AR3 is connected to the adjustment terminal of the adjustable resistor RP1 and is connected to the power supply terminal of the threshold power supply circuit through a resistor R5; the inverting input terminal of the operational amplifier AR4 is connected to the other end of the adjustable resistor RP1 and is grounded through a resistor R6; the inverting input terminal of the operational amplifier AR3 and the in-phase input terminal of the operational amplifier AR4 are connected to the output terminal of the operational amplifier AR2 through a resistor R4; the output terminals of the operational amplifier AR3 and the operational amplifier AR4 are connected to the anode of the transmitting terminal of the photoelectric coupler U1 through diodes VD3 and VD4 respectively, the cathode of the transmitting terminal of the photoelectric coupler U1 is grounded, the collector of the receiving terminal of the photoelectric coupler U1 is connected to the DCS control module and is connected to the power supply VCC through a resistor R11, and the emitter of the receiving terminal of the photoelectric coupler U1 is grounded.
5. The power DCS system fault monitoring device based on deep learning according to claim 4 is characterized in that: The threshold power supply circuit includes a +12V power supply, which is connected to the anode of the voltage zener diode DZ2 and one end of the adjustable resistor RP2 through a resistor R7, the cathode of the voltage zener diode DZ2 and the other end of the adjustable resistor RP2 are grounded, the adjustment end of the adjustable resistor RP2 is connected to the in-phase input end of the operational amplifier AR5 and one end of the resistor R9 through a resistor R8, the inverting input end of the operational amplifier AR5 is grounded through a resistor R10, and the output end of the operational amplifier AR5 is connected to the other end of the resistor R9 and the power supply end of the threshold power supply circuit.
6. The power DCS system fault monitoring device based on deep learning according to claim 1 is characterized in that: The data acquisition module also includes: A signal conditioning circuit, used for amplifying and harmonic suppression processing of the detection signal of the sensor array; The data acquisition card is used to perform analog-to-digital conversion on the signal processed by the signal conditioning circuit and transmit the converted digital signal to the DCS control module; the DCS control module manages the working state of the data acquisition card by sending control instructions.
7. The power DCS system fault monitoring device based on deep learning according to claim 6 is characterized in that: The signal conditioning circuit comprises: A differential amplification unit, used for performing differential amplification processing on the detection signal of the sensor array; A harmonic suppression unit, used for suppressing harmonic components in the differentially amplified signal; A dynamic compensation unit, used to collect the signal after harmonic suppression, and feed it back to the differential amplifier unit after processing through a compensation adjustment mechanism to eliminate signal fluctuations; The differential amplifier unit includes an operational amplifier AR6, an inverting input terminal of the operational amplifier AR6 is connected to the first detection signal input port through a resistor R21 and is grounded through a capacitor C11; a non-inverting input terminal of the operational amplifier AR6 is connected to the second detection signal input port through a resistor R22 and is grounded through a capacitor C12.
8. The power DCS system fault monitoring device based on deep learning according to claim 7 is characterized in that: The harmonic suppression unit includes an inductor L1, a capacitor C13, a capacitor C15 and a resistor R26. One end of the inductor L1 and the capacitor C13 is connected to the output end of the operational amplifier AR6, and the other end of the inductor L1 and the capacitor C13 is connected to one end of the resistor R26 and the data acquisition card. The other end of the resistor R26 is grounded through the capacitor C15.
9. The power DCS system fault monitoring device based on deep learning according to claim 8 is characterized in that: The dynamic compensation unit includes an operational amplifier AR7, an inverting input terminal of the operational amplifier AR7 is connected to one end of a resistor R26 via a resistor R25, and an output terminal of the operational amplifier AR7 is connected to a parallel resistor R24 and a capacitor C14, a non-inverting input terminal of the operational amplifier AR7 is grounded, and an output terminal of the operational amplifier AR7 is also connected to a non-inverting input terminal of an operational amplifier AR6 via a resistor R23.
10. The power DCS system fault monitoring device based on deep learning according to claim 1, characterized in that: The power management platform uses a deep learning model based on CNN+Bi-LSTM to perform real-time analysis of power data to achieve automatic diagnosis and early warning of faults.
Citation Information
Patent Citations
Method and system for monitoring and diagnosing power system
CN118094199A
