High-precision voltage detection method and detection system thereof

Abstract

The invention belongs to the technical field of voltage measurement, and particularly relates to a high-precision voltage detection method and system, and the method comprises the steps: carrying out the high-voltage detection of target equipment, and carrying out the collection and conversion of a high-voltage signal in the target equipment through a mutual inductor collection circuit; the collected signals are processed, the signals are transmitted to a signal modulation circuit through a filter network and an amplification circuit, and direct current signals in direct proportion to input sine wave signals are obtained through a low-pass filter; performing anomaly detection on the processed signal, determining and evaluating an abnormal voltage in real time by constructing a detection model, and comparing a difference value between prediction data and actual data to obtain a detection result; according to the invention, by constructing the detection model, the abnormal voltage can be judged and evaluated in real time, the difference value between the prediction data and the actual data can be compared, and the detection result can be obtained, so that the voltage stability of the target equipment can be ensured.

Description

Technical Field

[0001] The present invention belongs to the technical field of voltage measurement, and specifically relates to a high-precision voltage detection method and a detection system thereof. Background Art

[0002] As test instruments become increasingly integrated, producibility and maintainability have become important design features of test instruments. High-precision voltage detection technology can be applied to automatic calibration and testing of complete instruments, which can effectively improve the production and operation efficiency of complete instruments and reduce instrument R&D and maintenance costs.

[0003] The test instrument is composed of various functional module circuits, and there are many types of power supplies, which brings challenges to the reliability design of the test instrument. The working state of the power supply directly affects whether the test instrument can work normally, so the power supply voltage detection of each functional module is very important. When the test instrument is in an abnormal state and the voltage needs to be detected for fault diagnosis, it depends on manual troubleshooting by engineering and technical personnel. In addition to high requirements on the quality of technical personnel, there are also problems such as slow fault diagnosis speed, poor consistency, and manual operation to select the appropriate range when measuring voltages of different amplitudes. Summary of the invention

[0004] In order to partially or completely solve the above technical problems, the present invention provides a high-precision voltage detection method and a detection system thereof. The specific technical solutions are as follows:

[0005] In a first aspect, the present invention provides a method for high-precision voltage detection, comprising:

[0006] Step 1: Perform high voltage detection on the target device, and collect and convert the high voltage signal in the target device through the transformer collection circuit;

[0007] Step 2: Process the collected signal, transmit the signal to the signal modulation circuit through the filter network and the amplifier circuit, and obtain a DC signal proportional to the input sine wave signal through the low-pass filter;

[0008] Step 3: Perform anomaly detection on the processed signal, build a detection model, determine and evaluate the abnormal voltage in real time, compare the difference between the predicted data and the actual data, and obtain the detection result;

[0009] Step 4: Calibrate and synchronize the voltage signal value of the converted output signal.

[0010] Optionally, in step 1, high voltage is collected from the target device through an acquisition circuit. First, the voltage is measured through the input, and the corresponding current is output through the voltage sensor. The current is converted into a corresponding voltage through the measurement circuit, and then amplified through a voltage follower. The amplified signal is zeroed and compensated to output the acquisition signal.

[0011] Optionally, the signal processing in step 1 is to transmit the signal input in step 1 to a signal modulation circuit through a filtering network and an amplifying circuit, and obtain a DC signal proportional to the input sinusoidal wave signal through a low-pass filter. Finally, the data is sent to a single-chip microcomputer for data processing through an analog / digital converter and displayed on an LCD.

[0012] Optionally, in step 3, the signal is detected for abnormality by packaging the signal in step 2 into a data set and establishing a detection model, thereby inputting the voltage data set to be detected, obtaining the prediction result according to the detection rule, calculating the difference between the actual voltage and the detection voltage, and determining whether the target voltage is abnormal.

[0013] Optionally, a detection model is established to establish a model for detecting abnormal distribution network voltage data, and the steps are as follows:

[0014] Step 31, divide the historical voltage data into a training group and a test group, the test group is the voltage data to be detected; use the training set method to train the network voltage data, and use it to build a detection model;

[0015] Step 32: Count multiple voltage errors and calculate the voltage error standard deviation V, that is:

[0016]

[0017] In the formula, m represents the number of error values; e t It indicates the number of voltage errors occurring at time t; E indicates the standard value of the error;

[0018] Step 33: Based on the standard deviation value, a distribution network voltage prediction model is constructed:

[0019]

[0020] Where M represents voltage quality; w j It represents the j-th voltage feature distribution set. By using the distribution network voltage prediction model constructed above, the distribution network voltage abnormal data set can be obtained.

[0021] Optionally, after the detection model is established, the distribution network voltage prediction model is trained using normalization processing to improve the convergence speed of the voltage data set, that is:

[0022]

[0023] Where U t Indicates the actual voltage value at time t; U max Indicates the maximum voltage value in the detection sequence, U min Indicates the minimum voltage value in the detection sequence;

[0024] A set of data is generated from a series of normalized voltage sequences for training a recurrent neural network. The data set consists of multiple data lines, each of which consists of an input value and a corresponding label. Three consecutive sequence values ​​are taken as a straight line and labeled. The obtained data is reclassified into a training group and a test group. The network is trained by the training set method, tested and evaluated, and the average deviation between the predicted value and the labeled value of the neural network output layer is used as the loss function f1 for training, that is:

[0025]

[0026] In the formula, Represents the predicted value; n represents the number of samples in the training set. After the model training is completed, its performance is evaluated.

[0027] Optionally, the anomaly detection process in step 3 is as follows:

[0028] Step 311: Analyze the grid overload data in the distribution network data set according to the abnormal data identification characteristics, and use the initial value of the voltage data quality exceeding twice the rated value as the measurement mutation threshold. When the voltage to be detected is higher than the threshold, it indicates that the voltage is abnormal.

[0029] Step 312: by detecting multiple nodes with topological relationships, the voltage amplitude of each node is obtained. When the voltage amplitude reaches a certain accuracy, the corresponding distribution network voltage data can be detected and corrected;

[0030] Step 313: For multiple lines connected to the detection node, there is a maximum possible error in the voltage calculation value at that moment. If the terminal node is less than or equal to the maximum possible error, it means that the data in the detection result is normal data; otherwise, it means that the data in the detection result is an abnormal value.

[0031] In a second aspect, the present invention provides a high-precision voltage detection system for implementing any one of the high-precision voltage detection methods described in the first aspect, the detection system comprising: a voltage acquisition module, a data processing module, an abnormality detection module and an output adjustment module;

[0032] The voltage acquisition module performs high voltage detection on the target device and collects and converts the high voltage signal in the target device through the transformer acquisition circuit;

[0033] The data processing module processes the collected signal and transmits the signal to the signal modulation circuit through the filtering network and the amplifying circuit. After passing through the low-pass filter, a DC signal proportional to the input sine wave signal is obtained;

[0034] The anomaly detection module performs anomaly detection on the processed signal, builds a detection model, determines and evaluates the abnormal voltage in real time, compares the difference between the predicted data and the actual data, and obtains the detection result;

[0035] The output adjustment module calibrates and synchronizes the voltage signal value of the converted output signal.

[0036] Beneficial effects: The present invention performs high-voltage detection on the target device, collects and converts the high-voltage signal in the target device through a mutual inductor acquisition circuit; processes the collected signal, transmits the signal to a signal modulation circuit through a filter network and an amplification circuit, and obtains a DC signal proportional to the input sinusoidal wave signal after passing through a low-pass filter; performs abnormality detection on the processed signal, and determines and evaluates the abnormal voltage in real time by building a detection model, compares the difference between the predicted data and the actual data, and obtains the detection result; calibrates and synchronizes the voltage signal value of the converted output signal. The present invention can determine and evaluate the abnormal voltage in real time by building a detection model, compares the difference between the predicted data and the actual data, and obtains the detection result, thereby ensuring the voltage stability of the target device. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 It is a step diagram of the high-precision voltage detection method applied by the present invention.

[0038] Figure 2 It is a signal processing schematic diagram of the voltage high-precision detection method applied by the present invention.

[0039] Figure 3 It is a schematic diagram of establishing a detection model for the high-precision voltage detection method applied in the present invention.

[0040] Figure 4 It is a schematic diagram of abnormal detection of the voltage high-precision detection method applied by the present invention.

[0041] Figure 5 It is a schematic diagram of a high-precision voltage detection system applied for by the present invention.

[0042] Figure 6 A schematic diagram of the voltage signal value calibration of the present invention. DETAILED DESCRIPTION

[0043] The technical solution of the present invention will be described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments in 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.

[0044] like Figures 1 to 4As shown, in the first aspect, the present invention provides a method for high-precision voltage detection, comprising the following steps:

[0045] Step 1: Perform high voltage detection on the target device, and collect and convert the high voltage signal in the target device through the transformer collection circuit;

[0046] Step 2: Process the collected signal, transmit the signal to the signal modulation circuit through the filter network and the amplifier circuit, and obtain a DC signal proportional to the input sine wave signal through the low-pass filter;

[0047] Step 3: Perform anomaly detection on the processed signal. By building a detection model, the abnormal voltage is judged and evaluated in real time, and the difference between the predicted data and the actual data is compared to obtain the detection result.

[0048] Step 4: Calibrate and synchronize the voltage signal value of the converted output signal.

[0049] In some embodiments, Figures 1 to 4 As shown, in step 1, high voltage is collected from the target device through the collection circuit. First, the voltage is measured through the input, and the corresponding current is output through the voltage sensor. The current is converted into the corresponding voltage through the measurement circuit, and then amplified through the voltage follower. The amplified signal is zeroed and compensated, so as to output the collection signal.

[0050] In some embodiments, Figures 1 to 4 As shown, the signal processing in step 1 is to transmit the signal input in step 1 to the signal modulation circuit through the filtering network and the amplifying circuit, and obtain a DC signal proportional to the input sine wave signal through the low-pass filter, and finally send the data to the single-chip microcomputer for data processing through the analog / digital converter and display it on the LCD.

[0051] In some embodiments, Figures 1 to 4 As shown, in step 3, the signal abnormality detection is performed by packaging the signal in step 2 into a data set, and establishing a detection model, thereby inputting the voltage data set to be detected, obtaining the prediction result according to the detection rule, calculating the difference between the actual voltage and the detection voltage, and determining whether the target voltage is abnormal.

[0052] In some embodiments, Figures 1 to 4 As shown in the figure, a detection model is established to establish a model for detecting abnormal distribution network voltage data. The steps are as follows:

[0053] Step 31, divide the historical voltage data into a training group and a test group, the test group is the voltage data to be detected; use the training set method to train the network voltage data, and use it to build a detection model;

[0054] Step 32: Count multiple voltage errors and calculate the voltage error standard deviation V, that is:

[0055]

[0056] In the formula, m represents the number of error values; e t It indicates the number of voltage errors occurring at time t; E indicates the standard value of the error;

[0057] Step 33: Based on the standard deviation value, a distribution network voltage prediction model is constructed:

[0058]

[0059] Where M represents voltage quality; w j It represents the j-th voltage feature distribution set. By using the distribution network voltage prediction model constructed above, the distribution network voltage abnormal data set can be obtained.

[0060] In some embodiments, Figures 1 to 4 As shown in the figure, after the detection model is established, the distribution network voltage prediction model is trained using normalization processing to improve the convergence speed of the voltage data set, that is:

[0061]

[0062] Where U t Indicates the actual voltage value at time t; U max Indicates the maximum voltage value in the detection sequence, U min Indicates the minimum voltage value in the detection sequence;

[0063] A set of data is generated from a series of normalized voltage sequences for training a recurrent neural network. The data set consists of multiple data lines, each of which consists of an input value and a corresponding label. Three consecutive sequence values ​​are taken as a straight line and labeled. The obtained data is reclassified into a training group and a test group. The network is trained by the training set method, tested and evaluated, and the average deviation between the predicted value and the labeled value of the neural network output layer is used as the loss function f1 for training, that is:

[0064]

[0065] In the formula, Represents the predicted value; n represents the number of samples in the training set. After the model training is completed, its performance is evaluated.

[0066] In some embodiments, a large number of overvoltage data sets are automatically iterated by parameters, so there will be a part of abnormal waveforms in a large number of data waveforms, so the data is labeled and all time-frequency images are stored in JPEG format; the image is labeled according to the abnormal data annotation features and saved in the annotation, and any image in the JPEG image folder is trained through the XML document generated by each image and label, so that any XML document name in the annotation remains consistent.

[0067] In some embodiments, Figures 1 to 4 As shown, the anomaly detection process in step 3 is as follows:

[0068] Step 311: Analyze the grid overload data in the distribution network data set according to the abnormal data identification characteristics, and use the initial value of the voltage data quality exceeding twice the rated value as the measurement mutation threshold. When the voltage to be detected is higher than the threshold, it indicates that the voltage is abnormal.

[0069] Step 312: by detecting multiple nodes with topological relationships, the voltage amplitude of each node is obtained. When the voltage amplitude reaches a certain accuracy, the corresponding distribution network voltage data can be detected and corrected;

[0070] Step 313: For multiple lines connected to the detection node, there is a maximum possible error in the voltage calculation value at that moment. If the terminal node is less than or equal to the maximum possible error, it means that the data in the detection result is normal data; otherwise, it means that the data in the detection result is an abnormal value.

[0071] In some embodiments, Figure 6 As shown, in step 4, the voltage signal value calibration is to convert the voltage signal into a temperature signal, and compare the converted temperature signal with the threshold temperature signal through a comparison module. When the two temperature signals are not equal, the output signal is not zero, and the output signal is converted into a current of a DC circuit through an adjustment module, and the current of the DC circuit is changed, thereby changing the temperature signal of the DC quantity, until the temperature generated by the AC quantity is equal to the temperature generated by the DC quantity. According to the definition of the effective value, it can be known that the DC voltage at this time is the effective value of the detected AC quantity.

[0072] In a second aspect, the present invention provides a high-precision voltage detection system, such as Figure 5 As shown, a detection system for implementing a high-precision voltage detection method described in the first aspect includes: a voltage acquisition module, a data processing module, an abnormality detection module and an output adjustment module.

[0073] The voltage acquisition module performs high-voltage detection on the target device and collects and converts the high-voltage signal in the target device through the mutual inductor acquisition circuit; the data processing module processes the collected signal and transmits the signal to the signal modulation circuit through the filtering network and the amplification circuit. After passing through the low-pass filter, a DC signal proportional to the input sinusoidal wave signal is obtained; the anomaly detection module performs anomaly detection on the processed signal, and through the construction of a detection model, the abnormal voltage is judged and evaluated in real time, and the difference between the predicted data and the actual data is compared to obtain the detection result; the output adjustment module calibrates and synchronizes the voltage signal value of the converted output signal.

[0074] Optionally, in the anomaly detection module, establishing a detection model is used to establish a model for detecting abnormal data of distribution network voltage. The steps of establishing the detection model are as follows:

[0075] Step 31, divide the historical voltage data into a training group and a test group, the test group is the voltage data to be detected; use the training set method to train the network voltage data, and use it to build a detection model;

[0076] Step 32: Count multiple voltage errors and calculate the voltage error standard deviation V, that is:

[0077]

[0078] In the formula, m represents the number of error values; e t It indicates the number of voltage errors occurring at time t; E indicates the standard value of the error;

[0079] Step 33: Based on the standard deviation value, a distribution network voltage prediction model is constructed:

[0080]

[0081] Where M represents voltage quality; w j It represents the j-th voltage feature distribution set. By using the distribution network voltage prediction model constructed above, the distribution network voltage abnormal data set can be obtained.

[0082] Optionally, after the detection model is established, the distribution network voltage prediction model is trained using normalization processing to improve the convergence speed of the voltage data set, that is:

[0083]

[0084] Where U t Indicates the actual voltage value at time t; U max Indicates the maximum voltage value in the detection sequence, U min Indicates the minimum voltage value in the detection sequence;

[0085] A set of data is generated from a series of normalized voltage sequences for training a recurrent neural network. The data set consists of multiple data lines, each of which consists of an input value and a corresponding label. Three consecutive sequence values ​​are taken as a straight line and labeled. The obtained data is reclassified into a training group and a test group. The network is trained by the training set method, tested and evaluated, and the average deviation between the predicted value and the labeled value of the neural network output layer is used as the loss function f1 for training, that is:

[0086]

[0087] In the formula, Represents the predicted value; n represents the number of samples in the training set. After the model training is completed, its performance is evaluated.

[0088] In some embodiments, a large number of overvoltage data sets are automatically iterated by parameters, so there will be a part of abnormal waveforms in a large number of data waveforms, so the data is labeled and all time-frequency images are stored in JPEG format; the image is labeled according to the abnormal data annotation features and saved in the annotation, and any image in the JPEG image folder is trained through the XML document generated by each image and label, so that any XML document name in the annotation remains consistent.

[0089] According to the abnormal data identification characteristics, the power grid overload data in the distribution network data set is analyzed, and the initial value of the voltage data quality that is more than twice the rated value is used as the measurement mutation threshold. When the voltage to be detected is higher than the threshold, it means that the voltage is abnormal. By detecting multiple nodes with topological relationships, the voltage amplitude of each node is obtained. When the voltage amplitude reaches a certain accuracy, the corresponding distribution network voltage data can be detected and corrected. For multiple lines connected to the detection node, there is a maximum possible error in the voltage calculation value at that moment. If the terminal node is less than or equal to the maximum possible error, it means that the data in the detection result is normal data; otherwise, it means that the data in the detection result is an abnormal value.

[0090] In the present invention application, high voltage detection is performed on the target device, and the high voltage signal in the target device is collected and converted through the mutual inductor acquisition circuit; the collected signal is processed, and the signal is transmitted to the signal modulation circuit through the filtering network and the amplifying circuit, and a DC signal proportional to the input sinusoidal wave signal is obtained after passing through the low-pass filter; the processed signal is detected for abnormality, and by constructing a detection model, the abnormal voltage is judged and evaluated in real time, and the difference between the predicted data and the actual data is compared to obtain the detection result; the voltage signal value of the converted output signal is calibrated and synchronized. By constructing a detection model, the present invention application can judge and evaluate the abnormal voltage in real time, and compare the difference between the predicted data and the actual data to obtain the detection result, thereby ensuring the voltage stability of the target device.

[0091] Obviously, the above embodiments are merely examples for the purpose of clear explanation, and are not intended to limit the implementation methods. For those skilled in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to list all the implementation methods here. The obvious changes or modifications derived therefrom are still within the scope of protection created by the present invention.

Claims

1. A high-precision voltage detection method, characterized in that: include: Step 1: Perform high voltage detection on the target device, and collect and convert the high voltage signal in the target device through the transformer collection circuit; Step 2: Process the collected signal, transmit the signal to the signal modulation circuit through the filter network and the amplifier circuit, and obtain a DC signal proportional to the input sine wave signal through the low-pass filter; Step 3: Perform anomaly detection on the processed signal, build a detection model, determine and evaluate the abnormal voltage in real time, compare the difference between the predicted data and the actual data, and obtain the detection result; Step 4: Calibrate and synchronize the voltage signal value of the converted output signal.

2. A high-precision voltage detection method according to claim 1, characterized in that: In step 1, high voltage is collected from the target device through the collection circuit. First, the voltage is measured through the input, and the corresponding current is output through the voltage sensor. The current is converted into the corresponding voltage through the measurement circuit, and then amplified through the voltage follower. The amplified signal is zeroed and compensated to output the collection signal.

3. A high-precision voltage detection method according to claim 1, characterized in that: The signal processing in step 2 is to transmit the signal input in step 1 to the signal modulation circuit through the filtering network and the amplifying circuit, and obtain a DC signal proportional to the input sine wave signal through the low-pass filter. Finally, the data is sent to the single-chip microcomputer for data processing through the analog / digital converter and displayed on the LCD.

4. A high-precision voltage detection method according to claim 3, characterized in that: In step 3, the signal is detected for abnormality by packaging the signal in step 2 into a data set and establishing a detection model, thereby inputting the voltage data set to be detected, obtaining the prediction result according to the detection rule, calculating the difference between the actual voltage and the detection voltage, and determining whether the target voltage is abnormal.

5. A high-precision voltage detection method according to claim 4, characterized in that: Establishing the detection model A model for detecting abnormal voltage data in the distribution network was established. The steps are as follows: Step 31, divide the historical voltage data into a training group and a test group, the test group is the voltage data to be detected; use the training set method to train the network voltage data, and use it to build a detection model; Step 32: Count multiple voltage errors and calculate the voltage error standard deviation V, that is: In the formula, m represents the number of error values; e t It indicates the number of voltage errors occurring at time t; E indicates the standard value of the error; Step 33: Based on the standard deviation value, a distribution network voltage prediction model is constructed: Where M represents voltage quality; w j It represents the j-th voltage feature distribution set. By using the distribution network voltage prediction model constructed above, the distribution network voltage abnormal data set can be obtained.

6. A high-precision voltage detection method according to claim 5, characterized in that: After the detection model is established, the distribution network voltage prediction model is trained using normalization processing to improve the convergence speed of the voltage data set, namely: Where U t Indicates the actual voltage value at time t; U max Indicates the maximum voltage value in the detection sequence, U min Indicates the minimum voltage value in the detection sequence; A set of data is generated from a series of normalized voltage sequences for training a recurrent neural network. The data set consists of multiple data lines, each of which consists of an input value and a corresponding label. Three consecutive sequence values ​​are taken as a straight line and labeled. The obtained data is reclassified into a training group and a test group. The network is trained by the training set method, tested and evaluated, and the average deviation between the predicted value and the labeled value of the neural network output layer is used as the loss function f1 for training, that is: In the formula, Represents the predicted value; n represents the number of samples in the training set. After the model training is completed, its performance is evaluated.

7. A high-precision voltage detection method according to claim 6, characterized in that: The anomaly detection process in step 3 is as follows: Step 311: Analyze the grid overload data in the distribution network data set according to the abnormal data identification characteristics, and use the initial value of the voltage data quality exceeding twice the rated value as the measurement mutation threshold. When the voltage to be detected is higher than the threshold, it indicates that the voltage is abnormal. Step 312: by detecting multiple nodes with topological relationships, the voltage amplitude of each node is obtained. When the voltage amplitude reaches a certain accuracy, the corresponding distribution network voltage data can be detected and corrected; Step 313: For multiple lines connected to the detection node, there is a maximum possible error of the voltage calculation value at that moment. If the terminal node is less than or equal to the maximum possible error, it means that the data in the detection result is normal data; Otherwise, it means that the data in the test result is an abnormal value.

8. A high-precision voltage detection system, used to implement a high-precision voltage detection method as described in any one of claims 1 to 7, characterized in that: The detection system includes: a voltage acquisition module, a data processing module, an abnormality detection module and an output adjustment module; The voltage acquisition module performs high voltage detection on the target device and collects and converts the high voltage signal in the target device through the transformer acquisition circuit; The data processing module processes the collected signal and transmits the signal to the signal modulation circuit through the filtering network and the amplifying circuit. After passing through the low-pass filter, a DC signal proportional to the input sine wave signal is obtained; The anomaly detection module performs anomaly detection on the processed signal, builds a detection model, determines and evaluates the abnormal voltage in real time, compares the difference between the predicted data and the actual data, and obtains the detection result; The output adjustment module calibrates and synchronizes the voltage signal value of the converted output signal.

Images