A self-adaptive segmentation algorithm SF6 sensor system and alarm method

Abstract

This application relates to an SF6 sensor system and alarm method based on an adaptive segmentation algorithm, belonging to the technical field of gas detection. It includes an infrared sampling module, a matrix gain conditioning module, a signal state determination module, a dynamic gain feedback module, an adaptive segmentation calculation module, and a graded alarm output module. The system acquires SF6 infrared signals through measurement and reference channels. Based on the channel ratio, signal attenuation, analog-to-digital converter output state, and initial concentration estimation results, it determines the detection state, dynamically updates the hardware gain, software digital gain, and segment number, and calls the corresponding segment fitting model for continuous correction to obtain the true SF6 concentration value. This application has the effects of improving the sensitivity of trace leak detection, reducing the risk of high-concentration signal overflow, and achieving continuous output of graded alarms.

Description

Technical Field

[0001] This application relates to gas detection, and more particularly to an SF6 sensor system and alarm method based on an adaptive segmentation algorithm. Background Technology

[0002] Currently, SF6 gas is a key insulating and arc-quenching medium in power equipment, and its leakage detection is crucial for the safe operation of the power grid. Infrared absorption spectroscopy, due to its high selectivity and fast response speed, has become the mainstream technology for SF6 concentration detection. It typically employs a dual-channel structure with a measurement channel and a reference channel, using the Beer-Lambert law to invert the gas concentration. However, in practical applications, SF6 concentrations range widely, from trace leaks at the ppm level to sudden high concentrations at the volume percentage level during faults, posing a conflicting requirement for the sensor's detection range and accuracy.

[0003] The gain parameters of existing infrared SF6 sensors are usually fixed at the factory. High gain is beneficial for improving the sensitivity of trace detection, but it can lead to signal saturation and overflow in high-concentration scenarios. Low gain, while covering the high-concentration detection range, sacrifices the resolution and accuracy of trace detection. Some solutions use manual range switching or multi-segment calibration for compensation, but there are measurement blind spots or output jumps during range switching, and the concentration readings are discontinuous at the segment boundaries, affecting the reliability of the monitoring system. In addition, existing solutions lack a mechanism to automatically determine the current concentration range based on the real-time signal status and dynamically adjust the gain, making it impossible to achieve adaptive seamless detection from trace to high concentration ranges. In scenarios with rapid concentration changes, inaccurate detection or response lag is likely to occur.

[0004] Regarding the aforementioned technologies, existing SF6 sensor systems suffer from problems such as fixed gain leading to insufficient detection accuracy across the entire concentration range, discontinuous output during segment switching, and a lack of adaptive state determination and dynamic gain adjustment mechanisms, making it difficult to meet the application requirements of simultaneously addressing trace leakage early warning and high-concentration fault detection.

[0005] Therefore, this invention proposes an SF6 sensor system and alarm method with an adaptive segmentation algorithm. Summary of the Invention

[0006] The purpose of this application is to provide an SF6 sensor system and alarm method based on an adaptive segmentation algorithm.

[0007] To achieve the above objectives, this application adopts the following technical solution:

[0008] An SF6 sensor system with an adaptive segmentation algorithm includes the following modules:

[0009] The infrared sampling module is used to acquire the measurement channel signal corresponding to the characteristic absorption wavelength of SF6 and the reference channel signal corresponding to the non-absorption reference wavelength through an infrared light source, a gas chamber, a measurement channel, and a reference channel.

[0010] The matrix gain conditioning module is used to perform hardware gain conditioning and analog-to-digital conversion on the measurement channel signal and the reference channel signal according to the hardware gain in the current gain state, and to perform software digital compensation on the converted digital signal to obtain a dual-channel digital signal. The current gain state includes hardware gain, software digital gain and segment number.

[0011] The signal state determination module is used to calculate the channel ratio and signal attenuation based on the dual-channel digital signal, and combine the analog-to-digital conversion saturation state and the initial concentration estimation result to determine whether the current detection state is a trace detection state, a regular detection state, or a high-concentration anti-overflow state.

[0012] The dynamic gain feedback module is used to update the hardware gain, software digital gain and segment number in the current gain state according to the current detection state, and make the updated gain state take effect in the next sampling period.

[0013] The adaptive segmented calculation module is used to call the segmented fitting model according to the corresponding segment number within the sampling period when the adjusted gain state is in effect, and to continuously correct the fitting output results of adjacent segment boundaries to obtain the continuously output true SF6 concentration value.

[0014] The graded alarm output module is used to output a trace warning signal, a standard alarm signal, or a sudden danger alarm signal based on the actual SF6 concentration value.

[0015] Preferably, the infrared sampling module includes:

[0016] An infrared light source is used to emit infrared light containing the characteristic absorption wavelength of SF6 into the gas to be tested in the gas chamber;

[0017] The gas chamber is used to form the infrared absorption light path for the gas to be measured.

[0018] Measurement channel, used to acquire measurement channel signals through a 10.55μm narrowband filter;

[0019] The reference channel is used to acquire the reference channel signal through a 3.95μm or 3.91μm narrowband filter.

[0020] Preferably, the matrix gain conditioning module includes:

[0021] The analog gating unit is used to select the measurement channel signal and the reference channel signal to enter the gain conditioning path;

[0022] Programmable gain amplifiers and analog-to-digital converters are used to amplify, condition, and convert digital signals according to the hardware gain in the current gain state.

[0023] The software digital compensation unit is used to perform digital compensation according to the software digital gain in the current gain state;

[0024] The current gain state is determined by a gain state matrix, where each gain state row in the gain state matrix corresponds to a hardware gain, a software digital gain, and a segment number.

[0025] Preferably, in the signal state determination module, determining the current detection state includes:

[0026] The channel ratio is calculated based on the dual-channel digital signal, and the signal attenuation is determined based on the change of the channel ratio relative to the reference channel ratio.

[0027] The initial concentration estimation result is obtained based on the signal attenuation, and the state is determined in combination with the output state of the analog-to-digital converter.

[0028] When the analog-to-digital converter output is not in an overflow critical state, if the initial concentration estimation result is lower than the first set threshold, it is determined to be a trace detection state; if the initial concentration estimation result is between the first set threshold and the second set threshold, it is determined to be a normal detection state.

[0029] When the initial concentration estimate is higher than the second set threshold, or when the analog-to-digital converter output is in an overflow critical state, it is determined to be a high concentration overflow prevention state.

[0030] Preferably, in the dynamic gain feedback module, adjusting the hardware gain and software digital gain according to the current detection state includes:

[0031] In micro-detection mode, improve hardware gain and enhance software digital gain compensation;

[0032] Under normal testing conditions, conventional hardware gain and conventional software digital gain compensation are used;

[0033] Under high-concentration overflow prevention conditions, reduce hardware gain and decrease software digital gain compensation;

[0034] The adjusted hardware gain and software digital gain take effect in the next sampling period.

[0035] Preferably, in the adaptive segmented calculation module, the continuity correction includes:

[0036] The corresponding segment fitting model is called based on the currently effective hardware gain, software digital gain, and segment number to obtain the segment concentration calculation value;

[0037] When the segment numbering changes, the continuity correction is obtained by subtracting the concentration output value of the segment fitting model at the same segment boundary after the change from the concentration output value of the segment fitting model at the boundary of the adjacent segment before the change.

[0038] The continuous correction amount is superimposed on the segmented concentration calculation value after switching to obtain the continuously output true SF6 concentration value.

[0039] Preferably, in the graded alarm output module, outputting an alarm signal based on the actual SF6 concentration value includes:

[0040] Set progressively increasing thresholds for minor warnings, standard alarms, and sudden danger alarms;

[0041] When the actual SF6 concentration value is greater than the trace warning threshold but less than the standard alarm threshold, a trace warning signal is output.

[0042] When the actual SF6 concentration reaches the standard alarm threshold but is less than the sudden danger alarm threshold, a standard alarm signal is output.

[0043] When the actual SF6 concentration reaches the emergency alarm threshold, or when the current detection state is a high-concentration overflow prevention state and the actual SF6 concentration continues to rise, an emergency alarm signal will be output.

[0044] An adaptive segmentation algorithm-based alarm method for SF6 sensors includes the following steps:

[0045] S1. Acquire the measurement channel signal corresponding to the characteristic absorption wavelength of SF6 and the reference channel signal corresponding to the non-absorption reference wavelength, and perform hardware gain conditioning, analog-to-digital conversion and software digital compensation according to the current gain state to obtain dual-channel digital signals;

[0046] S2. Calculate the channel ratio and signal attenuation based on the dual-channel digital signal, and obtain the initial concentration estimation result based on the signal attenuation.

[0047] S3. Based on the initial concentration estimation result and the output status of the analog-to-digital converter, determine whether the current detection status is a trace detection status, a regular detection status, or a high-concentration anti-overflow status, and generate a gain status instruction for the next sampling cycle based on the current detection status.

[0048] S4. After the gain state command of the next sampling period takes effect, the corresponding piecewise fitting model is called to perform concentration fitting calculation, and the concentration output at the segment switching point is continuously corrected to obtain the true SF6 concentration value.

[0049] S5. Output a trace warning signal, a standard alarm signal, or a sudden danger alarm signal based on the actual SF6 concentration value.

[0050] Preferably, in step S3, determining the current detection state and generating a gain state instruction for the next sampling period based on the current detection state includes:

[0051] When the analog-to-digital converter output is not in an overflow critical state, and the initial concentration estimation result is lower than the first set threshold, it is determined to be a trace detection state, and a gain state instruction to improve hardware gain and enhance software digital gain compensation is generated.

[0052] When the analog-to-digital converter output is not in an overflow critical state, and the initial concentration estimation result is between the first set threshold and the second set threshold, it is determined to be in a normal detection state, and a gain state command for normal hardware gain and normal software digital gain compensation is generated.

[0053] When the initial concentration estimate is higher than the second set threshold, or when the analog-to-digital converter output is in an overflow critical state, it is determined to be a high concentration overflow prevention state, and a gain state instruction to reduce hardware gain and reduce software digital gain compensation is generated.

[0054] Preferably, in step S4, the continuity correction includes:

[0055] The current segment number is determined based on the gain status after it takes effect, and the corresponding segment fitting model is called to obtain the segment concentration calculation value;

[0056] When the current segment number is inconsistent with the segment number of the previous sampling period, calculate the concentration output difference between the two segment fitting models before and after the switch at the boundary of adjacent segments;

[0057] The concentration output difference is superimposed on the segmented concentration calculation value to obtain the true SF6 concentration value.

[0058] In summary, this application includes at least one of the following beneficial technical effects:

[0059] 1. By using the matrix gain conditioning module and the dynamic gain feedback module, the hardware gain and software digital gain are updated in real time according to the detection status. In the case of micro-level detection, the gain enhancement sensitivity is improved, and in the case of high-concentration anti-overflow, the gain extension linear range is reduced, so as to achieve adaptive detection across the entire concentration range.

[0060] 2. During segment switching, the adaptive segmented calculation module calculates the concentration difference at the boundary of the segmented fitting model before and after the switch as a continuity correction value and adds it to eliminate the concentration jump caused by segment switching and ensure continuous and smooth output.

[0061] 3. The detection status is determined by the signal status determination module based on four dimensions: channel ratio, signal attenuation, saturation state, and initial concentration estimation. Gain adjustment takes effect in the next sampling cycle to avoid timing conflicts between gain adjustment and signal acquisition in the same cycle, thus ensuring signal integrity and system stability. Attached Figure Description

[0062] Figure 1 This is a system framework diagram of the present invention;

[0063] Figure 2 This is a flowchart of the matrix gain feedback and detection state determination in this invention;

[0064] Figure 3 This is a flowchart of the adaptive segmented calculation and hierarchical alarm output in this invention;

[0065] Figure 4 This is a flowchart of the method of the present invention. Detailed Implementation

[0066] The following is in conjunction with the appendix Figure 1 -Appendix Figure 4 This application will be described in further detail below.

[0067] Example 1: An SF6 sensor system with an adaptive segmentation algorithm, referring to... Figure 1 ,include: The infrared sampling module is used to acquire the measurement channel signal corresponding to the characteristic absorption wavelength of SF6 and the reference channel signal corresponding to the non-absorption reference wavelength through an infrared light source, a gas chamber, a measurement channel, and a reference channel. The matrix gain conditioning module is used to perform hardware gain conditioning and analog-to-digital conversion on the measurement channel signal and the reference channel signal according to the hardware gain in the current gain state, and to perform software digital compensation on the converted digital signal to obtain a dual-channel digital signal. The current gain state includes hardware gain, software digital gain and segment number. The signal state determination module is used to calculate the channel ratio and signal attenuation based on the dual-channel digital signal, and combine the analog-to-digital conversion saturation state and the initial concentration estimation result to determine whether the current detection state is a trace detection state, a regular detection state, or a high-concentration anti-overflow state. The dynamic gain feedback module is used to update the hardware gain, software digital gain and segment number in the current gain state according to the current detection state, and make the updated gain state take effect in the next sampling period. The adaptive segmented calculation module is used to call the segmented fitting model according to the corresponding segment number within the sampling period when the updated gain state takes effect, and to continuously correct the fitting output results of adjacent segment boundaries to obtain the continuously output true SF6 concentration values. The graded alarm output module is used to output a trace warning signal, a standard alarm signal, or a sudden danger alarm signal based on the actual SF6 concentration value.

[0068] In one specific embodiment, the infrared sampling module includes an infrared light source, a gas chamber, a measurement channel, and a reference channel.

[0069] The infrared light source employs a MEMS broadband infrared light source to emit infrared light containing the characteristic absorption wavelength of SF6 into the gas to be tested within the gas chamber. The gas chamber can be an open reflective chamber with a high-reflectivity coating on its inner wall to increase the effective optical path of the infrared light within the chamber and improve the detection sensitivity for low-concentration SF6 gas.

[0070] The measurement channel is equipped with a 10.55μm narrowband filter to acquire the measurement channel signal corresponding to the characteristic absorption wavelength of SF6. The reference channel is equipped with a 3.95μm or 3.91μm narrowband filter to acquire the reference channel signal corresponding to the non-absorption reference wavelength. The measurement channel reflects the absorption changes of SF6 gas to characteristic wavelength infrared light, while the reference channel compensates for the effects of infrared light source fluctuations, gas chamber contamination, and environmental disturbances on the detection results.

[0071] The analog signal obtained from the sampling measurement channel is denoted as The analog signal of the reference channel is denoted as .

[0072] The matrix gain conditioning module includes an analog gating unit, a programmable gain amplifier, an analog-to-digital converter, and a software digital compensation unit.

[0073] The analog gating unit connects the measurement channel signal and the reference channel signal to the gain conditioning path respectively. The programmable gain amplifier amplifies and conditions the input signal according to the hardware gain in the current gain state. The analog-to-digital converter converts the amplified and conditioned signal into a digital signal. The software digital compensation unit performs digital compensation on the converted digital signal to obtain a dual-channel digital signal.

[0074] In this embodiment, the current gain state is determined by the gain state matrix. The gain state matrix can be represented as: ; in, For the i-th gain state line, For hardware gain, For software digital gain, For segment numbering.

[0075] The gain state matrix can include multiple gain state rows, for example: ; in, For the micro-detection state, a higher hardware gain and a corresponding software digital gain are used; For routine testing conditions, conventional hardware gain and conventional software digital gain are used. To prevent overflow in high-concentration states, a lower hardware gain and a smaller software digital gain are used.

[0076] For example, the hardware gain can be set to one or more combinations of 1x, 8x, 64x, or 512x. The specific gain level is not a limitation and can be determined by calibration based on the sensor chamber length, light source intensity, detector sensitivity, and ADC input range.

[0077] During system power-on initialization, the main control microprocessor sets the current gain state to the gain state line corresponding to the normal detection state, enabling the infrared sampling module and matrix gain conditioning module to first acquire and process data for the first sampling cycle according to the normal hardware gain, normal software digital gain, and normal segment numbering. Subsequently, the dynamic gain feedback module updates the gain state for the next sampling cycle based on the state determination results of each sampling cycle.

[0078] The digital signals of the measurement channel and the reference channel after software digital compensation are denoted as follows: .

[0079] like Figure 2 As shown, the signal state determination module calculates the channel ratio based on the dual-channel digital signal and determines the signal attenuation based on the change in the channel ratio relative to the reference channel ratio.

[0080] The channel ratio can be expressed as:

[0081] in, The channel ratio for the kth sampling period. For the digital signal of the measurement channel in the kth sampling period, This is the reference channel digital signal for the kth sampling period.

[0082] The system obtains the reference channel ratio in clean air or a calibration environment. The signal attenuation is determined based on the change in the current channel ratio relative to the reference channel ratio. The signal attenuation can be calculated using the following method:

[0083] in, The signal attenuation amount in the k-th sampling period, This is the baseline channel ratio.

[0084] The initial concentration estimation result is obtained based on the signal attenuation:

[0085] in, This is the initial concentration estimation result. This is the initial estimation function obtained through pre-calibration.

[0086] The signal state determination module also determines whether the analog-to-digital converter (ADC) output is in an overflow critical state. An overflow critical state can include: the ADC output approaching its upper or lower limit, or a trend over multiple consecutive sampling periods indicating that it is about to exceed the effective detection range. Specifically, an upper limit threshold can be set. and lower limit critical value .

[0087] When the analog-to-digital converter output is not in an overflow critical state, When the level is below the first set threshold, it is determined to be a trace detection state; when When the value is between the first and second preset thresholds, it is determined to be in a normal detection state; when... When the concentration exceeds the second set threshold, or when the analog-to-digital converter output is in an overflow critical state, it is determined to be a high-concentration overflow prevention state.

[0088] For example, the first threshold can be set to 30 ppm, and the second threshold can be set to 1500 ppm. These thresholds can also be reset according to the actual application scenario and calibration data.

[0089] The dynamic gain feedback module updates the hardware gain, software digital gain, and segment number in the current gain state based on the current detection state.

[0090] When the current detection state is a trace detection state, the dynamic gain feedback module generates a gain state command with higher hardware gain and stronger software digital gain compensation to improve the resolution of low-concentration signals.

[0091] When the current detection state is the normal detection state, the dynamic gain feedback module generates gain state instructions for normal hardware gain and normal software digital gain compensation in order to maintain stable detection within the normal concentration range.

[0092] When the current detection state is a high-concentration overflow prevention state, the dynamic gain feedback module generates a gain state command with lower hardware gain and smaller software digital gain compensation to expand the effective input range of the analog-to-digital conversion front end and reduce the risk of signal overflow during a sudden high-concentration leak.

[0093] To avoid the state determination result and gain adjustment result within the current sampling period from interfering with each other, this embodiment adopts a mechanism that takes effect in the next sampling period. That is, the gain state command generated in the k-th sampling period does not change the data processing result of the current sampling period, but takes effect at the beginning of the (k+1)-th sampling period.

[0094] The adaptive segmented calculation module calls the corresponding segmented fitting model based on the currently effective hardware gain, software digital gain, and segment number to obtain the segmented concentration calculation value.

[0095] The piecewise fitting model can be expressed as:

[0096] in, This represents the calculated segmented concentration value obtained in the p-th segment during the k-th sampling period. Let be the fitting function corresponding to the p-th segment. This represents the signal attenuation.

[0097] The piecewise fitting model is obtained by calibrating SF6 standard gas of known concentration under different gain states. During calibration, the relationship between signal attenuation and corresponding concentration value is recorded, and the fitting function corresponding to each segment number is formed by polynomial fitting, table lookup interpolation, or a combination of both.

[0098] To avoid abrupt changes in concentration output when switching between different gain states or different segmented models, this embodiment performs continuous correction. When the segment number changes from... Switch to At that time, the continuity correction is obtained by subtracting the concentration output value of the piecewise fitting model at the same segment boundary from the concentration output value of the piecewise fitting model at the boundary of adjacent segments before the switch.

[0099] in, For continuous correction amount, This represents the signal attenuation at the boundary between adjacent segments. To determine the concentration output value at the boundary of the piecewise fitting model before switching, This represents the concentration output value at the same boundary of the piecewise fitting model after switching.

[0100] The actual SF6 concentration value after the switch can be expressed as:

[0101] in, This is the true SF6 concentration value after continuous correction. These are the segmented concentration calculation values ​​obtained from the segmented fitting model after the switch.

[0102] When the segment numbering has not changed, it can be set to The current segment concentration calculation value is directly used as the actual SF6 concentration value.

[0103] like Figure 3 As shown, the graded alarm output module outputs alarm signals based on the actual SF6 concentration value. The system is set with progressively increasing thresholds: trace warning threshold, standard alarm threshold, and sudden danger alarm threshold, denoted as [reference needed]. , and .

[0104] When the actual SF6 concentration value is greater than the trace warning threshold but less than the standard alarm threshold, a trace warning signal is output to indicate the presence of early leakage risk.

[0105] When the actual SF6 concentration reaches the standard alarm threshold but is less than the sudden danger alarm threshold, a standard alarm signal is output. The standard alarm signal can be output to an audible and visual alarm, controller, or ventilation linkage device.

[0106] When the actual SF6 concentration reaches the emergency alarm threshold, or when the current detection status is high-concentration overflow prevention and the actual SF6 concentration continues to rise, an emergency alarm signal is output. This emergency alarm signal can be output to a hazard linkage terminal for use in conjunction with security systems, emergency shut-off devices, or remote dispatch platforms.

[0107] In one embodiment, the trace warning threshold, standard alarm threshold, and sudden danger alarm threshold can be set according to the application scenario and equipment calibration requirements. For example, the trace warning threshold can be lower than 1000 ppm, the standard alarm threshold can be set to 1000 ppm, and the sudden danger alarm threshold can be set to 1500 ppm.

[0108] It should be noted that the first and second set thresholds are used for detection state determination and gain state selection, while the micro-early warning threshold, standard alarm threshold, and sudden danger alarm threshold are used for alarm level output. The two can be set separately according to calibration results and application scenarios.

[0109] Example 2: This example also provides an SF6 sensor alarm method with an adaptive segmentation algorithm, which includes the following steps.

[0110] S1. Acquire the measurement channel signal corresponding to the characteristic absorption wavelength of SF6 and the reference channel signal corresponding to the non-absorption reference wavelength, and perform hardware gain conditioning, analog-to-digital conversion and software digital compensation according to the current gain state to obtain dual-channel digital signals.

[0111] S2. Calculate the channel ratio and signal attenuation based on the dual-channel digital signal, and obtain the initial concentration estimation result based on the signal attenuation.

[0112] S3. Based on the initial concentration estimation results and the output status of the analog-to-digital converter, determine whether the current detection status is a trace detection status, a regular detection status, or a high-concentration anti-overflow status, and generate a gain status instruction for the next sampling cycle based on the current detection status.

[0113] S4. After the gain state command of the next sampling period takes effect, the corresponding piecewise fitting model is called to perform concentration fitting calculation, and the concentration output at the segment switching point is continuously corrected to obtain the true SF6 concentration value.

[0114] S5. Output a trace warning signal, a standard alarm signal, or a sudden danger alarm signal based on the actual SF6 concentration value.

[0115] Using the above methods, the system can increase the gain in the low concentration range to enhance the detection capability of trace leaks, and decrease the gain in the high concentration range to reduce the risk of signal overflow. It can also output continuous real SF6 concentration values ​​through piecewise fitting models and continuous correction.

[0116] The embodiments described in this specific implementation are preferred embodiments of this application and are not intended to limit the scope of protection of this application. Identical components are represented by the same reference numerals. Therefore, all equivalent changes made to the structure, shape, and principle of this application should be covered within the scope of protection of this application.

Claims

1. An SF6 sensor system with an adaptive segmentation algorithm, comprising the following modules, characterized in that: The infrared sampling module is used to acquire the measurement channel signal corresponding to the characteristic absorption wavelength of SF6 and the reference channel signal corresponding to the non-absorption reference wavelength through an infrared light source, a gas chamber, a measurement channel, and a reference channel. The matrix gain conditioning module is used to perform hardware gain conditioning and analog-to-digital conversion on the measurement channel signal and the reference channel signal according to the hardware gain in the current gain state, and to perform software digital compensation on the converted digital signal to obtain a dual-channel digital signal. The current gain state includes hardware gain, software digital gain and segment number. The signal state determination module is used to calculate the channel ratio and signal attenuation based on the dual-channel digital signal, and combine the analog-to-digital conversion saturation state and the initial concentration estimation result to determine whether the current detection state is a trace detection state, a regular detection state, or a high-concentration anti-overflow state. The dynamic gain feedback module is used to update the hardware gain, software digital gain and segment number in the current gain state according to the current detection state, and make the updated gain state take effect in the next sampling period. The adaptive segmented calculation module is used to call the segmented fitting model according to the corresponding segment number within the sampling period when the adjusted gain state is in effect, and to continuously correct the fitting output results of adjacent segment boundaries to obtain the continuously output true SF6 concentration value. The graded alarm output module is used to output a trace warning signal, a standard alarm signal, or a sudden danger alarm signal based on the actual SF6 concentration value.

2. The SF6 sensor system with an adaptive segmentation algorithm according to claim 1, characterized in that, The infrared sampling module includes: An infrared light source is used to emit infrared light containing the characteristic absorption wavelength of SF6 into the gas to be tested in the gas chamber; The gas chamber is used to form the infrared absorption light path for the gas to be measured. Measurement channel, used to acquire measurement channel signals through a 10.55μm narrowband filter; The reference channel is used to acquire the reference channel signal through a 3.95μm or 3.91μm narrowband filter.

3. The SF6 sensor system with an adaptive segmentation algorithm according to claim 1, characterized in that, The matrix gain conditioning module includes: The analog gating unit is used to select the measurement channel signal and the reference channel signal to enter the gain conditioning path; Programmable gain amplifiers and analog-to-digital converters are used to amplify, condition, and convert digital signals according to the hardware gain in the current gain state. The software digital compensation unit is used to perform digital compensation according to the software digital gain in the current gain state; The current gain state is determined by a gain state matrix, where each gain state row in the gain state matrix corresponds to a hardware gain, a software digital gain, and a segment number.

4. The SF6 sensor system with an adaptive segmentation algorithm according to claim 1, characterized in that, In the signal state determination module, determining the current detection state includes: The channel ratio is calculated based on the dual-channel digital signal, and the signal attenuation is determined based on the change of the channel ratio relative to the reference channel ratio. The initial concentration estimation result is obtained based on the signal attenuation, and the state is determined in combination with the output state of the analog-to-digital converter. When the analog-to-digital converter output is not in an overflow critical state, if the initial concentration estimation result is lower than the first set threshold, it is determined to be a trace detection state; if the initial concentration estimation result is between the first set threshold and the second set threshold, it is determined to be a normal detection state. When the initial concentration estimate is higher than the second set threshold, or when the analog-to-digital converter output is in an overflow critical state, it is determined to be a high concentration overflow prevention state.

5. The SF6 sensor system with an adaptive segmentation algorithm according to claim 1, characterized in that, In the dynamic gain feedback module, adjusting the hardware gain and software digital gain according to the current detection state includes: In micro-detection mode, improve hardware gain and enhance software digital gain compensation; Under normal testing conditions, conventional hardware gain and conventional software digital gain compensation are used; Under high-concentration overflow prevention conditions, reduce hardware gain and decrease software digital gain compensation; The adjusted hardware gain and software digital gain take effect in the next sampling period.

6. The SF6 sensor system with an adaptive segmentation algorithm according to claim 1, characterized in that, In the adaptive segmented calculation module, continuity correction includes: The corresponding segment fitting model is called based on the currently effective hardware gain, software digital gain, and segment number to obtain the segment concentration calculation value; When the segment numbering changes, the continuity correction is obtained by subtracting the concentration output value of the segment fitting model at the same segment boundary after the change from the concentration output value of the segment fitting model at the boundary of the adjacent segment before the change. The continuous correction amount is superimposed on the segmented concentration calculation value after switching to obtain the continuously output true SF6 concentration value.

7. The SF6 sensor system with an adaptive segmentation algorithm according to claim 1, characterized in that, The graded alarm output module outputs an alarm signal based on the actual SF6 concentration value, including: Set progressively increasing thresholds for minor warnings, standard alarms, and sudden danger alarms; When the actual SF6 concentration value is greater than the trace warning threshold but less than the standard alarm threshold, a trace warning signal is output. When the actual SF6 concentration reaches the standard alarm threshold but is less than the sudden danger alarm threshold, a standard alarm signal is output. When the actual SF6 concentration reaches the emergency alarm threshold, or when the current detection state is a high-concentration overflow prevention state and the actual SF6 concentration continues to rise, an emergency alarm signal will be output.

8. An SF6 sensor alarm method based on an adaptive segmentation algorithm, used in the SF6 sensor system based on the adaptive segmentation algorithm described in claims 1-7, characterized in that, Includes the following steps: S1. Acquire the measurement channel signal corresponding to the characteristic absorption wavelength of SF6 and the reference channel signal corresponding to the non-absorption reference wavelength, and perform hardware gain conditioning, analog-to-digital conversion and software digital compensation according to the current gain state to obtain dual-channel digital signals; S2. Calculate the channel ratio and signal attenuation based on the dual-channel digital signal, and obtain the initial concentration estimation result based on the signal attenuation. S3. Based on the initial concentration estimation result and the output status of the analog-to-digital converter, determine whether the current detection status is a trace detection status, a regular detection status, or a high-concentration anti-overflow status, and generate a gain status instruction for the next sampling cycle based on the current detection status. S4. After the gain state command of the next sampling period takes effect, the corresponding piecewise fitting model is called to perform concentration fitting calculation, and the concentration output at the segment switching point is continuously corrected to obtain the true SF6 concentration value. S5. Output a trace warning signal, a standard alarm signal, or a sudden danger alarm signal based on the actual SF6 concentration value.

9. The SF6 sensor alarm method based on an adaptive segmentation algorithm according to claim 8, characterized in that, In step S3, determining the current detection state and generating a gain state instruction for the next sampling period based on the current detection state includes: When the analog-to-digital converter output is not in an overflow critical state, and the initial concentration estimation result is lower than the first set threshold, it is determined to be a trace detection state, and a gain state instruction to improve hardware gain and enhance software digital gain compensation is generated. When the analog-to-digital converter output is not in an overflow critical state, and the initial concentration estimation result is between the first set threshold and the second set threshold, it is determined to be in a normal detection state, and a gain state command for normal hardware gain and normal software digital gain compensation is generated. When the initial concentration estimate is higher than the second set threshold, or when the analog-to-digital converter output is in an overflow critical state, it is determined to be a high concentration overflow prevention state, and a gain state instruction to reduce hardware gain and reduce software digital gain compensation is generated.

10. The SF6 sensor alarm method based on an adaptive segmentation algorithm according to claim 8, characterized in that, In step S4, the continuity correction includes: The current segment number is determined based on the gain status after it takes effect, and the corresponding segment fitting model is called to obtain the segment concentration calculation value; When the current segment number is inconsistent with the segment number of the previous sampling period, calculate the concentration output difference between the two segment fitting models before and after the switch at the boundary of adjacent segments; The concentration output difference is superimposed on the segmented concentration calculation value to obtain the true SF6 concentration value.

Images