A Blind Source Separation Gas CO and H2 Detection System Based on Enhanced Channel
By adopting a blind source separation technology with enhanced channels in CO and H2 field detection, CO and H2 electrochemical sensors are used side by side and signal separation, the problem of poor detection accuracy and accuracy in the prior art is solved, and a high-precision and stable consistent gas detection effect is achieved.
Patent Information
- Application Number
- CN202211545259.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-12-02
- Publication Date
- 2025-03-04
- Estimated Expiration
- 2042-12-02
AI Technical Summary
The existing CO and H2 on-site detection technologies have problems such as large equipment size, high testing environment requirements, long operating time, and high operator requirements. Electrochemical detection has problems such as cross-response and unstable response curves, resulting in poor detection accuracy and accuracy.
The blind source separation technology based on enhanced channel is adopted, and the superimposed signal with improved signal-to-noise ratio is obtained through parallel use of CO and H2 electrochemical sensors, and the blind source separation technology is used to optimize the separation and reconstruction of CO and H2 content signals.
It significantly improves detection accuracy, reduces environmental requirements and equipment performance stability requirements, and has high stability and consistency in gas detection accuracy.
Smart Images
Figure CN115856215B_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the field of substation operation and inspection, and in particular relates to a blind source separation gas CO and H2 detection system based on enhanced channels. Background Art
[0002] The construction of new power systems is accelerating, and GIS electrical equipment with excellent insulation and arc extinguishing performance is widely used. How to ensure the healthy operation of such equipment is a vital task for power safety production. The concentration of components such as CO can better reflect the operating status of the equipment, and is also an effective and fast means of current GIS on-site operation and maintenance detection. At the same time, by enhancing the characteristics of the transmission channel and thus improving the signal-to-noise ratio, the method is widely used and developed in the field of communications; by applying blind source separation technology, the aliased signal is effectively separated and the original information is reconstructed, which shines in the fields of radar and digital signal processing.
[0003] The existing on-site detection technologies for CO and H2 have the following shortcomings: 1) Chromatographic detection technology uses the difference in the adsorption coefficients of two phases to separate and detect the content of components. The equipment is large in size, has high requirements for the test environment, long operation time, and high requirements for operators. It is not suitable for on-site detection of CO and H2. 2) Infrared detection technology uses the different absorption coefficients of gas molecules to the energy spectrum to respond to and detect the content of components. The equipment is precise, not suitable for transportation, and has high requirements for test temperature and humidity. It is not suitable for on-site detection of CO and H2. 3) The current mainstream technology for on-site detection of CO and H2 in GIS is the electrochemical method. This method uses the electrochemical reaction characteristics of CO and H2 to achieve component response and detection. This method is simple and convenient, but due to the characteristics of electrochemistry itself and manufacturing processes, there are problems such as being affected by the on-site environment, large cross-response, and unstable response curves. The accuracy and precision of direct response value measurement are poor. Summary of the invention
[0004] The purpose of the present invention is to provide a blind source separation gas CO and H2 detection system based on enhanced channels, which adopts electrochemical technology suitable for on-site detection, and obtains a superimposed signal with improved signal-to-noise ratio by using CO and H2 electrochemical sensors in parallel. At the same time, its selectivity is utilized to overcome the uncertainty problem of the separation signal order of the blind source separation technology; then, the blind source separation technology is used to optimally separate and reconstruct the CO and H2 content signals.
[0005] To achieve the above object, the technical solution of the present invention is: a blind source separation gas CO, H2 detection system based on enhanced channel, including an enhanced acquisition unit and a blind source separation unit; wherein,
[0006] The enhanced acquisition unit includes a CO sensor and a H2 sensor, wherein the CO sensor and the H2 sensor are connected in parallel to the signal input terminal, and the CO response value output terminal of the CO sensor and the H2 response value output terminal of the H2 sensor are respectively connected to the system noise as the signal output terminal;
[0007] The blind source separation unit includes a mixed matrix construction module, a matrix centralization module, a natural gradient iterative solution module, and a detection signal reconstruction module which are connected in sequence. The input of the mixed matrix construction module is connected to the signal output end of the enhanced acquisition unit, and the output of the detection signal reconstruction module serves as the output end of the blind source separation unit, i.e., the detection value output end of CO and H2.
[0008] Compared with the prior art, the present invention has the following beneficial effects:
[0009] 1) Make full use of the useful information in the cross signal, overcome the simple treatment of the cross signal as noise, and significantly improve the detection accuracy;
[0010] 2) The sensor selectivity is used to solve the problem of signal order uncertainty in blind source separation, thus improving the usability of the technology when the source signal strengths are similar;
[0011] 3) Make full use of the insensitivity of blind source separation technology to the signal transmission / superposition process, reduce the environmental requirements for accurate testing of CO and H2 and the performance stability requirements of the testing equipment, and the stability and consistency of gas detection accuracy are high. BRIEF DESCRIPTION OF THE DRAWINGS
[0012] Figure 1 This is a block diagram of the system principle of the present invention.
[0013] Figure 2 This is the natural gradient iteration process of the present invention. DETAILED DESCRIPTION
[0014] The technical solution of the present invention is described in detail below in conjunction with the accompanying drawings.
[0015] The following are specific implementation examples of the present invention.
[0016] like Figure 1 As shown, the present invention provides a blind source separation gas CO and H2 detection system based on enhanced channel, and the system includes two parts: enhanced acquisition and blind source separation.
[0017] (1) Enhanced acquisition link. The enhanced acquisition node includes gas response sensor selection and sensor signal connection. When selecting gas response sensors, select the CO sensor with strong CO response and the H2 sensor with strong H2 response respectively. When connecting sensor signals, at the input end, the CO sensor and the H2 sensor are connected in parallel; at the output end, the H2 response value output end and the CO response value output end of the H2 sensor and the system noise are connected and output. According to the above selection and connection, the enhanced acquisition link inputs the aliased signal x of CO and H2, and outputs the detection signals y1 and y2.
[0018] (2) Blind source separation: The blind source separation node includes four steps: building a mixing matrix, matrix centering, natural gradient iterative solution, and detection signal reconstruction.
[0019] (2.1) Construct the mixing matrix and construct y1 and y2 into vectors Then construct the mixing matrix Among them, y11, y12, ... y1n: n discrete detection values output by the detection signal y1; y21, y22, ... y2n: n discrete detection values output by the detection signal y2.
[0020] (2.2) Matrix centralization step: the mixed matrix Centralization, get Where, average(Y): the average value of matrix Y.
[0021] (2.3) Natural gradient iterative solution link, the natural gradient iterative solution node includes two steps: whitening processing and CO and H2 gradient iterative processing, such as Figure 2 shown.
[0022] (2.3.1) Whitening process, input mixing matrix Y ave , output Z = W whitening *Y ave .
[0023] (2.3.2) CO and H2 gradient iteration processing stage, that is, processing two gradient iterations for CO and H2 components respectively. First initialize the gradient matrix W = Rand, check whether the gradient matrix gradient satisfies the convergence condition div(W) > Critical, if yes, complete the iteration and obtain the separation matrix W; if not, update the gradient matrix W i =mean(Z·tanh(W i-1 ·Z) 2 )·Z, and then iteratively calculate W i =W i-1 -div(W i-1 ) / W i-1 , normalized Then check whether the gradient matrix gradient satisfies the convergence condition div(W)>Critical. Among them, Wwhitening is the whitening matrix; Rand is the random matrix; Wi-1 is the i-1th iteration matrix, Wi is the i-th iteration matrix, and the gradient matrix of the successful iteration is the separation matrix W; tanh is the hyperbolic tangent function; div(W) is the gradient of W; Critical sets the empirical threshold.
[0024] (3) CO and H2 detection signal reconstruction step, obtaining the detection value of CO and H2 X = W T ·Z.
[0025] The present invention has the following advantages:
[0026] 1) Make full use of the useful information in the cross signal, overcome the simple treatment of the cross signal as noise, and significantly improve the detection accuracy;
[0027] 2) The sensor selectivity is used to solve the problem of signal order uncertainty in blind source separation, thus improving the usability of the technology when the source signal strengths are similar;
[0028] 3) Make full use of the insensitivity of blind source separation technology to the signal transmission / superposition process, reduce the environmental requirements for accurate testing of CO and H2 and the performance stability requirements of the testing equipment, and the stability and consistency of gas detection accuracy are high.
[0029] The above are preferred embodiments of the present invention. Any changes made according to the technical solution of the present invention, as long as the resulting functions do not exceed the scope of the technical solution of the present invention, belong to the protection scope of the present invention.
Claims
1. A blind source separation gas CO and H2 detection system based on enhanced channel, characterized in that: It includes an enhanced acquisition unit and a blind source separation unit; wherein, The enhanced acquisition unit includes a CO sensor and a H2 sensor, wherein the CO sensor and the H2 sensor are connected in parallel to the signal input terminal, and the CO response value output terminal of the CO sensor and the H2 response value output terminal of the H2 sensor are respectively connected to the system noise as the signal output terminal; The blind source separation unit includes a mixed matrix construction module, a matrix centralization module, a natural gradient iterative solution module, and a detection signal reconstruction module which are connected in sequence. The input of the mixed matrix construction module is connected to the signal output end of the enhanced acquisition unit. The output of the detection signal reconstruction module serves as the output end of the blind source separation unit, i.e., the detection value output end of CO and H2. The enhanced acquisition unit specifically implements the following functions: the signal input terminal inputs the aliased signal x of CO and H2, The detection signals y1 and y2 are outputted through the CO sensor with strong response to CO, the H2 sensor with strong response to H2, and the system noise respectively; The specific function of the mixed matrix construction module is to construct y1 and y2 into vectors. ,and then Construct the mixing matrix Y = ; Wherein, y11, y12, ...y1n: n discrete detection values output by the detection signal y1; y21, y22, ...y2n: n discrete detection values output by the detection signal y2; The matrix centralization module specifically implements the function of: converting the mixing matrix Y= Centralization, get = ; Wherein, average(Y): average value of matrix Y; The natural gradient iterative solution module implements two steps of whitening treatment and CO and H2 gradient iterative treatment, wherein the whitening treatment step inputs the mixing matrix , output ; CO and H2 gradient iteration processing steps, that is, processing two gradient iterations for CO and H2 components respectively, first initialize the gradient matrix , check whether the gradient matrix gradient meets the convergence condition If yes, the iteration is completed and the separation matrix W is obtained; if no, the gradient matrix is updated , and then iteratively calculate , normalized After that, check the gradient matrix gradient until the convergence condition is met. ; Wherein, Wwhitening is the whitening matrix; Rand is the random matrix; Wi-1 is the i-1th iteration matrix, Wi is the i-th iteration matrix, and the gradient matrix of the successful iteration is the separation matrix W; tanh is the hyperbolic tangent function; div(W) is the gradient of W; Critical sets the empirical threshold; The specific implementation function of the detection signal reconstruction module is: based on the separation matrix W, the detection values of CO and H2 are obtained .