A fault diagnosis method and system based on electrical box parameter prediction

Abstract

The present invention discloses a fault diagnosis method and system based on electrical box parameter prediction, specifically: 1: collecting actual parameters of the electrical box; 2: preprocessing the collected data; judging whether there are abnormal parameters based on the data collected at the current moment, if so, go to step 5; otherwise, go to step 3; 3: using a time series analysis method to construct an autoregressive sliding average model, and based on the collected data, using the autoregressive sliding average model to predict the parameters at the next moment; 4: judging whether there are abnormal parameters in the predicted parameters at the next moment, if so, go to step 5; otherwise, go to step 1; 5: extracting abnormal parameters, judging the fault level according to the abnormal parameters, and then constructing an abnormal parameter matrix, and standardizing the data in the abnormal parameter matrix; 6: performing feature extraction on the abnormal parameter matrix, so as to judge the fault type and location of the electrical box at the current moment, and issue an early warning based on the fault type.

Description

Technical Field

[0001] The present invention belongs to the technical field of electrical equipment, and in particular to a fault diagnosis method and system based on electrical box parameter prediction. Background Art

[0002] In modern industry, buildings, and power systems, electrical boxes are core equipment for power distribution and control. Their stable operation is crucial to the reliability and safety of the entire system. However, over long-term operation, electrical boxes are prone to various failures due to factors such as component aging, overload, and environmental factors (such as high temperature, humidity, and corrosive gases). These failures can even lead to serious accidents such as fires and equipment damage.

[0003] Traditional electrical box monitoring relies primarily on manual inspections, which present numerous limitations. For one thing, manual inspections occur at long intervals, making it difficult to capture abnormal changes in the electrical box's operation in real time and identify potential faults promptly. Furthermore, the accuracy of manual judgment is easily affected by the inspector's experience and subjective factors, leading to the risk of missed detections and misjudgments. Furthermore, some existing monitoring systems can only monitor a single or a few parameters (such as current and voltage), failing to fully reflect the operating status of the electrical box. Even some systems capable of multi-parameter monitoring often rely on simple threshold comparison methods for data processing and fault diagnosis. These methods lack the ability to accurately identify complex fault patterns, making it difficult to provide effective early warning and diagnosis of faults. These systems fail to meet the demands of modern power systems for high-reliability and intelligent O&M of electrical boxes. Summary of the Invention

[0004] Purpose of the invention: In order to solve the problems existing in the above-mentioned prior art, the present invention provides a fault diagnosis method and system based on electrical box parameter prediction.

[0005] Technical solution: The present invention discloses a fault diagnosis method based on electrical box parameter prediction, which specifically includes the following steps:

[0006] Step 1: Collect the actual parameters of the electrical box in real time;

[0007] Step 2: Preprocess the collected data and construct a multidimensional time series for any parameter after preprocessing , t represents time; based on the data collected at the current moment, determine whether there are abnormal parameters. If so, go to step 5; otherwise, go to step 3;

[0008] Step 3: Use the time series analysis method to build an autoregressive moving average model. Based on the collected data, use the autoregressive moving average model to predict the parameters at the next moment.

[0009] Step 4: Determine whether there are abnormal parameters in the parameters of the next moment to be predicted. If so, go to step 5; otherwise, go to step 1;

[0010] Step 5: Extract abnormal parameters, determine the fault level based on the abnormal parameters, then construct an abnormal parameter matrix and standardize the data in the abnormal parameter matrix;

[0011] Step 6: Extract features from the abnormal parameter matrix to determine the fault type and location of the electrical box at the current moment, and issue an early warning based on the fault type.

[0012] Furthermore, a corresponding first threshold is set for each parameter, and the abnormal parameter is a parameter whose value exceeds the corresponding first threshold.

[0013] Furthermore, the parameters of the electrical box in step 1 include electrical parameters and environmental parameters. The electrical parameters include current, voltage, active power, reactive power, apparent power and frequency. The environmental parameters include air humidity, temperature and gas composition inside the electrical box.

[0014] Furthermore, the preprocessing includes adaptive filtering and normalization.

[0015] Furthermore, the step 3 is specifically as follows:

[0016] Step 3.1: Based on the historical data of any parameter, establish the autoregressive moving average model of the parameter:

[0017] ;

[0018] in, is the autoregressive order, is the sliding average order, is the autoregressive coefficient, is the sliding average coefficient, is a white noise sequence, For the forgetting factor;

[0019] Step 3.2: Use the following formula to dynamically determine the optimal autoregressive order and the sliding average order :

[0020]

[0021] in, represents the Akaike Information Criterion value, ; For the likelihood function value, select The values ​​of p and q corresponding to the minimum value;

[0022] Step 3.3: Update the parameters of the model in step 3.1 in real time using the following formula:

[0023] ;

[0024] in, is the gain matrix, is the covariance matrix, is the parameter vector, , is the historical data vector, , T represents transpose;

[0025] Calculate the absolute value of the difference between the predicted value and the actual value of the parameter at time t ,like ; then the predicted value of the parameter at time t is interpolated; k is 2 or 3, is the residual standard deviation;

[0026] Step 3.4: Based on the model parameters at time t, predict the parameters of the electrical box at time t+1:

[0027] ;

[0028] in, A multidimensional time series consisting of the parameters to be predicted.

[0029] Furthermore, in step 5, the following formula is used to standardize the abnormal data:

[0030] ;

[0031] in, For the data after standardization, is the data in row a and column b of the abnormal parameter matrix, is the mean, is the standard deviation.

[0032] Furthermore, step 6 is specifically as follows:

[0033] Step 5.1: Use the CNN model to predict the probability of fault type and fault location. The convolution expression used by the CNN model is as follows:

[0034] ;

[0035] in, It is The output of the convolutional layer The first feature Rank Characteristics of the elements of the column, It is The convolutional layer The weight corresponding to the element in the mth row and nth column of the feature, It is The input features of the convolutional layer, is the bias term, is the size of the convolution kernel;

[0036] The pooling layer in the CNN model uses adaptive pooling;

[0037] Step 5.2: The fault type and fault location corresponding to the maximum probability output by the CNN model are taken as the final fault type and fault location.

[0038] Furthermore, the step 6 is to perform an early warning based on the fault type, specifically setting a corresponding first threshold, a second threshold and a third threshold for each parameter; for each abnormal parameter among all abnormal parameters, if the value of the abnormal parameter is greater than the corresponding first threshold Less than or equal to the corresponding second threshold ; The fault is considered to be a level 1 fault, and the warning light will turn orange. Indicates the The first threshold corresponding to the abnormal parameter, Indicates the The second threshold value corresponding to the abnormal parameters is f=1,2,…,F; F represents the total number of abnormal parameters. If there is an abnormal parameter The value is greater than the second threshold Less than or equal to the third threshold , the fault is considered a secondary fault, the warning light turns yellow, and the volume of 0 decibels is used to remind the staff. is the second threshold corresponding to the abnormal parameter Y, is the third threshold corresponding to the abnormal parameter Y; if there is an abnormal parameter The value is greater than the third threshold If the fault is determined to be a level 3 fault, the warning light will display red and a volume greater than Q decibels will be used to remind the working staff.

[0039] A fault diagnosis system based on electrical box parameter prediction, comprising:

[0040] Parameter monitoring module, used to monitor the actual parameters of the electrical box in real time and transmit them to the data processing module;

[0041] A data processing module, used for pre-processing the received data;

[0042] The data prediction module is used to build an autoregressive moving average model and predict the value of the parameter at the next moment based on the autoregressive moving average model;

[0043] Fault diagnosis module, used to extract abnormal parameters and determine the fault type and fault location based on the abnormal parameters;

[0044] Alarm module, used to issue alarm information according to the fault type;

[0045] The power module includes a power supply and a backup battery. When the power supply is normal, the power supply supplies power to the system and charges the backup battery. When the power supply fails, the backup battery provides power.

[0046] The remote monitoring module is used to transmit the system's operating status, monitoring data and fault diagnosis reports to the remote monitoring center in real time through network communication. Managers can access the remote monitoring center through terminal devices to view the operating status of the electrical box. Managers can also adjust parameters of the electrical box and restart the equipment through the remote monitoring module.

[0047] Beneficial effects: The present invention realizes multi-parameter real-time monitoring of the electrical parameters and environmental parameters of the electrical box, and can fully and accurately grasp the operating status of the electrical box. The adaptive filtering algorithm is used to process the data to effectively remove noise interference and improve the accuracy and reliability of the data. By establishing a change trend model of the operating parameters of the electrical box, the operating trend of the electrical box can be predicted and potential faults can be discovered in advance. The use of artificial intelligence algorithms for fault diagnosis improves the accuracy and efficiency of fault diagnosis, and can quickly and accurately predict the type, location and severity of the fault. It has multiple early warning methods, and can provide graded early warnings according to the severity of the fault, so that relevant personnel can be notified in time to facilitate the adoption of corresponding measures. The remote monitoring function allows managers to understand the operating status of the electrical box anytime and anywhere, and can perform remote operations, which improves the convenience and efficiency of management. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 This is a system block diagram of the present invention.

[0049] Figure 2 Flowchart of the present invention. DETAILED DESCRIPTION

[0050] The accompanying drawings, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0051] like Figure 1-2 As shown, this embodiment specifically includes the following parts:

[0052] Multi-parameter monitoring module: used to monitor the electrical parameters and environmental parameters of the electrical box. The electrical parameter monitoring unit includes a current sensor, a voltage sensor, a power sensor, and a frequency sensor installed inside the electrical box. The power sensor calculates the active power, reactive power, and apparent power of the electrical box based on the data measured by the current sensor and the voltage sensor. The frequency sensor monitors the fluctuation of the power supply frequency. The electrical parameter monitoring unit converts the collected electrical parameter information into electrical signals and transmits them to the data processing module via wired or wireless means. The environmental parameter monitoring unit includes a temperature sensor, a humidity sensor, and a gas sensor. The temperature sensor detects the temperature near the electronic components inside the electrical box. The humidity sensor measures the air humidity inside the electrical box. The gas sensor detects harmful gases generated inside the electrical box due to overheating of electrical components and decomposition of insulating materials. The environmental parameter monitoring unit transmits the collected data to the data processing module.

[0053] Data processing module: Receives electrical parameter and environmental parameter data transmitted by the multi-parameter monitoring module, filters the data to remove noise interference, and uses an adaptive filtering algorithm to automatically adjust the filtering parameters according to the real-time changes in the data. Performs real-time analysis and calculation on the processed data, determines whether the current operating status of the electrical box is normal based on the set normal operating parameter range, and stores and analyzes historical data. Establishes a trend model for the operating parameters of the electrical box. The specific method is: normalize the historical data, and for the g-th electrical parameter , the normalized formula is ; is the minimum value of the gth electrical parameter, is the maximum value of the gth electrical parameter, g=1,2,…,G; G represents the total category of electrical parameters; for environmental parameters , the normalized formula is ; Represents the normalized environmental parameters. represents the minimum value of the hth environmental parameter, is the maximum value of the hth environmental parameter, h=1,2,…,H; H represents the total category of environmental parameters.

[0054] Using time series analysis method, construct autoregressive moving average model , and estimate the model parameters by the least squares method to determine the optimal and Values ​​are used to establish a trend model of the electrical box operating parameters changing over time; specifically:

[0055] Obtain historical electrical parameters and environmental parameters from the data processing module and construct a multidimensional time series for any parameter , standardize the parameters of different dimensions: , is the mean, is the standard deviation; is the value after normalization.

[0056] Dynamically determine the optimal autoregressive order through AIC and the sliding average order ;

[0057] ;

[0058] in, ; is the likelihood function value.

[0059] right and Calculate AIC and select the one corresponding to the minimum value .

[0060] Introducing the forgetting factor , giving higher weight to recent data and suppressing the influence of historical noise, thus establishing a trend model of the electrical box operating parameters changing over time: .

[0061] in, is the autoregressive coefficient, is the sliding average coefficient, is the white noise term;

[0062] Use recursive least squares method to update parameters. The iterative formula is:

[0063] ;

[0064] in, is the parameter vector.

[0065] is the historical data vector.

[0066] is the gain matrix, is the covariance matrix;

[0067] Outlier detection and correction

[0068] Calculate model prediction residuals , Represents the parameters predicted at time t, if , Take 2 or 3, If is the residual standard deviation, the prediction parameter is considered an outlier; the outlier is replaced by linear interpolation of adjacent time data.

[0069] Trend forecasting and model updating

[0070] Based on the current parameters, predict the next moment value: ;

[0071] After collecting several new samples, the order optimization and parameter estimation are re-executed to adapt to the changes in the operating status.

[0072] The data processing module also performs pattern recognition on abnormal data through artificial intelligence algorithms and transmits it to the fault diagnosis module.

[0073] Fault diagnosis module: Receives abnormal data, determines the fault type, location, and severity, and generates a fault diagnosis report. The specific method is as follows:

[0074] Preprocess the abnormal data and record the abnormal data matrix as ,in is the sample size, Number of features; use standardization method to process data, for each feature , and its standardized formula is: ,in is the mean, is the standard deviation.

[0075] CNN is used for feature extraction and pattern recognition. Dilated convolution is introduced in the convolution layer to expand the receptive field. The expression is as follows:

[0076] ;

[0077] in, It is The output of the convolutional layer The first feature Rank Characteristics of the elements of the column, It is The convolutional layer Elements of characteristics The corresponding weight, It is The input features of the convolutional layer, is the bias term, is the size of the convolution kernel.

[0078] Adaptive pooling is used instead of fixed-size pooling to dynamically adjust the pooling window size according to the feature map; a GAP layer is added after the last convolution layer to reduce parameters and enhance translation invariance: ,in, is the feature map size, For the The global mean of the feature maps.

[0079] After being processed by the convolutional layer and the pooling layer, the feature map is flattened into a one-dimensional vector and input into the fully connected layer;

[0080] The fully connected layer uses multi-task learning to simultaneously predict fault types and location :

[0081]

[0082] in is the one-dimensional feature vector output by the pooling layer, is the number of input neurons.

[0083] Apply Softmax to each task branch separately and convert the output of the fully connected layer into a probability distribution:

[0084] ;

[0085] Output vector: ; represents the type probability vector, Represents the position probability vector, and Select the type and position corresponding to the largest probability.

[0086] Early warning module: According to the severity of the fault in the fault diagnosis report, different levels of early warning methods are adopted. Specifically, each parameter sets the corresponding first threshold, second threshold and third threshold; for each parameter among all abnormal parameters, if the value of the parameter is greater than the corresponding first threshold Less than or equal to the corresponding second threshold ; The fault is considered to be a level 1 fault, and the warning light will turn orange. Indicates the The first threshold corresponding to the abnormal parameter, Indicates the The second threshold value corresponding to the abnormal parameters is f=1,2,…,F; F represents the total number of abnormal parameters. If there is an abnormal parameter The value is greater than the second threshold Less than or equal to the third threshold , the fault is considered a secondary fault, the warning light turns yellow, and a volume of 1 decibel is used to alert the staff. is the second threshold corresponding to the abnormal parameter Y, is the third threshold corresponding to the abnormal parameter Y; if there is an abnormal parameter The value is greater than the third threshold If the fault is determined to be a level 3 fault, the warning light will show red and a volume greater than 1 decibel will be used to remind the staff.

[0087] Remote monitoring module: The system's operating status, monitoring data, and fault diagnosis reports are transmitted to the remote monitoring center in real time through network communication, allowing managers to access the remote monitoring center through terminal devices to view the operating status of the electrical box, and allowing managers to remotely adjust parameters of the electrical box and restart the equipment.

[0088] Power supply module: Provides stable power supply for the entire system. It adopts dual power supply, including AC power supply and backup battery power supply. When the AC power is normal, the AC power supplies the system and charges the backup battery. When the AC power fails, the backup battery automatically switches to power supply.

[0089] In addition, the current sensor adopts a closed-loop Hall current sensor; the voltage sensor adopts a resistive voltage divider voltage sensor; the temperature sensor adopts a digital temperature sensor; the gas sensor includes a carbon monoxide sensor and a hydrogen sensor, and adopts a semiconductor gas sensor.

[0090] The specific implementation process of this embodiment is as follows:

[0091] In practical applications, the sensors of the multi-parameter monitoring module are installed in appropriate locations within the electrical box. Current sensors (closed-loop Hall effect current sensors), voltage sensors (resistive voltage divider voltage sensors), power sensors, and frequency sensors collect real-time electrical parameters such as current, voltage, power, and frequency. Temperature sensors (digital temperature sensors), humidity sensors, and gas sensors (carbon monoxide and hydrogen sensors, both using semiconductor gas sensors) monitor environmental parameters such as temperature, humidity, and harmful gases within the box. The collected data is converted into electrical signals and transmitted to the data processing module via wired or wireless means.

[0092] After receiving the data, the data processing module uses an adaptive filtering algorithm to filter the data and remove noise interference. Then, according to the set normal operating parameter range, the processed data is analyzed and calculated in real time to determine whether the current operating status of the electrical box is normal. At the same time, the historical data is normalized and an autoregressive sliding average model is constructed using time series analysis methods. ,The model parameters are estimated by the least square method, and a ,changing trend model of the operating parameters of the electrical ,box is established.

[0093] When the data processing module detects abnormal data, it transmits it to the fault diagnosis module. This module preprocesses and standardizes the abnormal data, uses CNN for feature extraction and pattern recognition, and after processing through convolutional, pooling, and fully connected layers, uses the softmax function to convert the output into a probability distribution to determine the fault type, location, and severity. It then generates a fault diagnosis report.

[0094] The early warning module takes corresponding early warning measures according to the severity of the fault in the fault diagnosis report.

[0095] The remote monitoring module transmits the system's operating status, monitoring data, and fault diagnosis reports to a remote monitoring center in real time via network communications. Managers can access the remote monitoring center via mobile phones, computers, and other devices to view the electrical box's operating status and remotely adjust parameters and restart the box.

[0096] When the mains power is normal, the power module uses the mains power to power the system and charge the backup battery; when the mains power fails, the backup battery automatically switches to power supply to ensure the normal operation of the system.

[0097] It should also be noted that the various specific technical features described in the above specific embodiments can be combined in any appropriate manner without contradiction. To avoid unnecessary repetition, the present invention will not further describe various possible combinations.

Claims

1. A fault diagnosis method based on electrical box parameter prediction, characterized in that: The specific steps include: Step 1: Collect the actual parameters of the electrical box in real time; Step 2: Preprocess the collected data and construct a multidimensional time series for any parameter after preprocessing , t represents time; Based on the data collected at the current moment, determine whether there are abnormal parameters. If so, go to step 5; otherwise, go to step 3; Step 3: Use the time series analysis method to build an autoregressive moving average model. Based on the collected data, use the autoregressive moving average model to predict the parameters at the next moment. Step 4: Determine whether there are abnormal parameters in the parameters of the next moment to be predicted. If so, go to step 5; otherwise, go to step 1; Step 5: Extract abnormal parameters, determine the fault level based on the abnormal parameters, then construct an abnormal parameter matrix and standardize the data in the abnormal parameter matrix; Step 6: Extract features from the abnormal parameter matrix to determine the fault type and location of the electrical box at the current moment, and issue an early warning based on the fault type; Step 3.1: Based on the historical data of any parameter, establish the autoregressive moving average model of the parameter: ； in, is the autoregressive order, is the sliding average order, is the autoregressive coefficient, is the sliding average coefficient, is a white noise sequence, For the forgetting factor; Step 3.2: Use the following formula to dynamically determine the optimal autoregressive order and the sliding average order : ； in, represents the Akaike Information Criterion value, ; For the likelihood function value, select The minimum value corresponds to value; Step 3.3: Update the parameters of the model in step 3.1 in real time using the following formula: ； in, is the gain matrix, is the covariance matrix, is the parameter vector, , is the historical data vector, , T represents transpose; Calculate the absolute value of the difference between the predicted value and the actual value of the parameter at time t ,like ; Then the predicted value of the parameter at time t is interpolated; Take 2 or 3, is the residual standard deviation; Step 3.4: Based on the model parameters at time t, predict the parameters of the electrical box at time t+1: ； in, A multidimensional time series consisting of the parameters to be predicted; Step 6 uses the CNN model to predict the probability of fault type and fault location.

2. A fault diagnosis method based on electrical box parameter prediction according to claim 1, characterized in that: A corresponding first threshold is set for each parameter, and the abnormal parameter is a parameter whose value exceeds the corresponding first threshold.

3. A fault diagnosis method based on electrical box parameter prediction according to claim 1, characterized in that: The parameters of the electrical box in step 1 include electrical parameters and environmental parameters. The electrical parameters include current, voltage, active power, reactive power, apparent power and frequency. The environmental parameters include air humidity and gas composition inside the electrical box.

4. A fault diagnosis method based on electrical box parameter prediction according to claim 1, characterized in that: The preprocessing includes adaptive filtering and normalization.

5. The fault diagnosis method based on electrical box parameter prediction according to claim 1, characterized in that: In step 5, the following formula is used to standardize the abnormal data: ； in, For the data after standardization, is the data in row a and column b of the abnormal parameter matrix, is the mean, is the standard deviation.

6. A fault diagnosis method based on electrical box parameter prediction according to claim 1, characterized in that: Step 6 is as follows: Step 5.1: The expression of convolution used by the CNN model is as follows: ； in, It is The output of the convolutional layer The first feature Rank Characteristics of the elements of the column, It is The convolutional layer The weight corresponding to the element in the mth row and nth column of the feature, It is The input features of the convolutional layer, is the bias term, is the size of the convolution kernel; The pooling layer in the CNN model uses adaptive pooling; Step 5.2: The fault type and fault location corresponding to the maximum probability output by the CNN model are taken as the final fault type and fault location.

7. The fault diagnosis method based on electrical box parameter prediction according to claim 1, characterized in that: In step 6, the warning is performed based on the fault type. Specifically, each parameter sets a corresponding first threshold, a second threshold, and a third threshold; for each abnormal parameter among all abnormal parameters, if the value of the abnormal parameter is greater than the corresponding first threshold Less than or equal to the corresponding second threshold ; The fault is considered to be a level 1 fault, and the warning light will turn orange. Indicates the The first threshold corresponding to the abnormal parameter, Indicates the The second threshold value corresponding to the abnormal parameters is f=1,2,…,F; F represents the total number of abnormal parameters. If there is an abnormal parameter The value is greater than the second threshold Less than or equal to the third threshold , the fault is considered a secondary fault, the warning light turns yellow, and the volume of 0 decibels is used to remind the staff. is the second threshold corresponding to the abnormal parameter Y, is the third threshold corresponding to the abnormal parameter Y; if there is an abnormal parameter The value is greater than the third threshold If the fault is determined to be a level 3 fault, the warning light will display red and a volume greater than Q decibels will be used to remind the working staff.

8. Implementing the fault diagnosis method based on electrical box parameter prediction as described in claim 1, characterized in that: include: Parameter monitoring module, used to monitor the actual parameters of the electrical box in real time and transmit them to the data processing module; A data processing module, used for pre-processing the received data; The data prediction module is used to build an autoregressive moving average model and predict the value of the parameter at the next moment based on the autoregressive moving average model; Fault diagnosis module, used to extract abnormal parameters and determine the fault type and fault location based on the abnormal parameters; Alarm module, used to issue alarm information according to the fault type; The power module includes a power supply and a backup battery. When the power supply is normal, the power supply supplies power to the system and charges the backup battery. When the power supply fails, the backup battery provides power. The remote monitoring module is used to transmit the system's operating status, monitoring data and fault diagnosis reports to the remote monitoring center in real time through network communication. Managers can access the remote monitoring center through terminal devices to view the operating status of the electrical box. Managers can also adjust parameters of the electrical box and restart the equipment through the remote monitoring module.

Images