Detection method and system for indoor hydrogen leakage positioning

Abstract

The invention discloses a detection method and system for indoor hydrogen leakage localization, and the method comprises the steps: building a three-dimensional network of MEMS gas sensors indoors, and obtaining the coordinates of a node where each MEMS gas sensor is located; the method comprises the following steps: measuring response characteristics of indoor gas by using MEMS gas sensors, and carrying out baseline compensation by adopting an EEMD-GRNN ensemble empirical mode decomposition-generalized regression neural network model to obtain a time-varying curve of hydrogen concentration of each MEMS gas sensor; fitting a time-varying curve of the hydrogen concentration by using a Gaussian distribution solution of a Fick second law equation to obtain an optimal fitting parameter so as to obtain the distance between the MEMS gas sensor and the hydrogen leakage position; according to the distance between the MEMS gas sensor and the hydrogen leakage position, the occurrence position of indoor hydrogen leakage is obtained.

Description

Technical Field

[0001] The present invention relates to the technical field of indoor hydrogen leakage position identification, and in particular to a detection method and system for indoor hydrogen leakage positioning. Background Art

[0002] As an ideal clean energy, hydrogen energy is an important breakthrough in dealing with energy crises, environmental pollution and other issues. It has important application prospects in transportation, industry, power generation and other fields. Hydrogen is generally stored in hydrogen storage tanks and transported to the equipment through pipelines. However, during storage or use, it is easy to leak due to aging of valves, loose sealing and other reasons, which may cause serious consequences such as explosions. Therefore, during the storage, transportation and use of hydrogen, hydrogen leakage must be strictly monitored and the hydrogen leakage location must be located.

[0003] In the detection and location of hydrogen leaks, the acoustic wave method or the sensor method is usually used. However, the acoustic wave method is easily affected by the surrounding environmental noise and the mechanical vibration of the equipment when in use. At the same time, it is also limited by the detection frequency band. The narrow frequency is easy to miss some acoustic signals, and the wide frequency has weak anti-interference ability. As for the sensor method, mobile robots equipped with sensors are currently used to map gas distribution and locate leaks. For indoor environments with more and more complex equipment, the robot's movement is restricted and it is difficult to achieve full coverage.

[0004] The above information disclosed in this Background section is only for enhancement of understanding of the background of the invention and therefore it may contain information that does not form the prior art that is already known to a person of ordinary skill in the art. Summary of the invention

[0005] The present invention provides a detection method and system for indoor hydrogen leakage positioning, which achieves full coverage and high anti-interference and recognition accuracy.

[0006] Detection methods used to locate indoor hydrogen leaks include:

[0007] Step 1: Build a three-dimensional network of MEMS gas sensors indoors and obtain the coordinates (x, y, z) of each node where the MEMS gas sensor is located;

[0008] Step 2: Use MEMS gas sensors to measure the response characteristics of indoor gas, use EEMD-GRNN ensemble empirical mode decomposition-generalized regression neural network model to perform baseline compensation, and obtain the curve of hydrogen concentration change over time of each MEMS gas sensor;

[0009] Step 3: Use the Gaussian distribution solution of Fick's second law equation to fit the curve of the change of hydrogen concentration over time to obtain the best fitting parameters, so as to obtain the distance between the MEMS gas sensor and the hydrogen leakage position;

[0010] Step 4: Obtain the location of the indoor hydrogen leakage based on the distance between the MEMS gas sensor and the hydrogen leakage location.

[0011] In the detection method for indoor hydrogen leakage positioning, in step 1, the three-dimensional network of the MEMS gas sensor is composed of n identical MEMS gas sensors placed at different nodes indoors, n is an integer greater than or equal to 4, and a three-dimensional coordinate is established with the MEMS gas sensor on one of the nodes as the origin, and the coordinates (x, y, z) of the MEMS gas sensors at all other nodes are obtained.

[0012] In the detection method for indoor hydrogen leakage location, step 2 comprises:

[0013] Step 2.1, the MEMS gas sensor is selected as a sensor for detecting hydrogen as the target gas;

[0014] Step 2.2, calibrate the MEMS gas sensor, use the EEMD-GRNN ensemble empirical mode decomposition-generalized regression neural network model to denoise the baseline signal and then fit the baseline response signal, and compensate for the drift of the MEMS gas sensor response baseline by making a difference;

[0015] Step 2.21, calibrate the MEMS gas sensor, and use EEMD ensemble empirical decomposition to denoise the MEMS gas sensor response baseline signal;

[0016] Step 2.22, divide the denoised MEMS gas sensor response baseline signal into a training set and a validation set with a ratio of 4:1, and normalize the divided training data and validation data to reduce the amount of model calculation;

[0017] Step 2.23, set the radial basis function of the GRNN generalized regression neural network model to a Gaussian function, set the smoothing factor parameter, and use the training data set and the set smoothing factor parameter to build the GRNN generalized regression neural network model;

[0018] Step 2.24, use the GRNN generalized regression neural network model to predict the validation set data and calculate the mean square error between the predicted value and the true value;

[0019] Step 2.25, repeat steps 2.23-2.24, use different smoothing factor parameters, and calculate the mean square error between the predicted value and the true value;

[0020] Step 2.26, by comparing the mean square error between the predicted value and the true value obtained using different smoothing factor parameters, the optimal smoothing factor parameter is obtained, and a GRNN generalized regression neural network model is built to fit the denoised MEMS gas sensor response baseline signal;

[0021] Step 2.27, judging the fitting result by comparing the fitting curve with the original curve after denoising, increasing the smoothing factor when overfitting, and reducing the smoothing factor when underfitting; compensating for the drift of the MEMS gas sensor response baseline by making a difference;

[0022] Step 2.3, use the MEMS gas sensor to obtain the response signal to the target gas, and after cleaning the data using the missing value processing algorithm and the outlier processing algorithm, obtain the curve of the change of hydrogen concentration over time.

[0023] In the detection method for indoor hydrogen leakage location, in step 2.2,

[0024] Perform baseline calibration of MEMS gas sensor under air;

[0025] The response baseline change curve of the MEMS gas sensor over time is input into the EEMD-GRNN ensemble empirical mode decomposition-generalized regression neural network model, and the baseline signal is decomposed and recombined using EEMD to achieve the purpose of noise reduction, and the drift trend signal is obtained. The drift trend signal is modeled and fitted using the EEMD-GRNN ensemble empirical mode decomposition-generalized regression neural network model to obtain the prediction curve of the response baseline;

[0026] The predicted baseline is subtracted from the response curve of the MEMS gas sensor to hydrogen to complete the baseline compensation.

[0027] In the detection method for indoor hydrogen leakage location, in step 2.3,

[0028] Use MEMS gas sensors to detect unknown concentrations of hydrogen and obtain response signals;

[0029] Install q centralized processing units indoors, where q is an integer greater than or equal to 2, and reasonably connect the MEMS gas sensors to the centralized processing units according to the locations of the MEMS gas sensors and the centralized processing units; the centralized processing units acquire response signals measured by each sensor and transmit them wirelessly to a computer;

[0030] After baseline compensation of the response signal of the MEMS gas sensor, the missing value processing algorithm and the outlier processing algorithm are used to clean the data. For the missing values ​​in the response signal, first identify the missing data points, which can be achieved through the ismissing function in MATLAB, and calculate the average of the two data points around it for interpolation; the median absolute deviation method is used to process the outliers in the target gas response signal, calculate the absolute deviation between each data point and the median, and find the median of these deviations, set the threshold to 3 times the median deviation, and remove the data points that exceed the threshold; obtain the curve of hydrogen concentration change over time.

[0031] In the detection method for indoor hydrogen leakage location, step 3 comprises:

[0032] Step 3.1, the Gaussian distribution solution of Fick's second law equation is ,in is the concentration at a certain node in the room, is the hydrogen concentration at the point where the indoor hydrogen storage equipment leaks, r is the distance between the node and the hydrogen leakage location, D is the diffusion coefficient, and t is the diffusion time;

[0033] Step 3.2, using the Gaussian distribution solution to fit the curve of the hydrogen concentration changing with time, to obtain the best fitting parameters , D and r, thereby obtaining the distance r between the MEMS gas sensor and the indoor hydrogen leakage location;

[0034] Step 3.3, evaluate the fitting results, and evaluate the quality of the fitting results by calculating the residual, mean square error and root mean square error. Among them, the fitting accuracy can be improved by adjusting the model parameters of the EEMD-GRNN ensemble empirical mode decomposition-generalized regression neural network model or increasing the number of data points.

[0035] In the detection method for indoor hydrogen leakage location, step 4 includes,

[0036] Step 4.1, sort the MEMS gas sensors from small to large according to their distance r from the hydrogen leakage location;

[0037] Step 4.2, select four MEMS gas sensors that are closest to the hydrogen leak location and are not in the same plane as the positioning nodes of the three-side positioning method;

[0038] Step 4.3, draw spheres in ascending order with the positioning nodes as the sphere center and the distance r between them and the hydrogen leakage position as the radius. The first two spheres intersect to obtain a closed curve. The third sphere intersects with the closed curve obtained by the first two spheres to obtain two points. The point closest to the fourth sphere among these two points is the predicted position of the indoor hydrogen leakage position.

[0039] In the detection method for indoor hydrogen leakage location, hydrogen leakage area identification includes:

[0040] According to the arrangement position of the MEMS gas sensors, the maximum distance Q between the three closest MEMS gas sensors and the n MEMS gas sensors is obtained. n , where n is an integer greater than or equal to 4, the largest Q n The value is recorded as M;

[0041] With the maximum Q n The value M is a judgment standard. If the distance between a certain MEMS gas sensor and another MEMS gas sensor is less than or equal to M, then the two MEMS gas sensors are adjacent MEMS gas sensors to each other.

[0042] The MEMS gas sensor that first generates a response signal is selected as a first judgment reference, and the MEMS gas sensors that subsequently generate response signals are judged in turn to be adjacent MEMS gas sensors of the first judgment reference;

[0043] If the subsequent three MEMS gas sensors are all adjacent MEMS gas sensors of the first judgment criterion, there is a single hydrogen leakage area; if one of the subsequent three MEMS gas sensors is not an adjacent MEMS gas sensor of the first judgment criterion, this MEMS gas sensor is used as the second judgment criterion to judge whether the subsequent MEMS gas sensors are adjacent MEMS gas sensors of the first judgment criterion or the second judgment criterion, until three adjacent MEMS gas sensors of the second judgment criterion appear, if the middle MEMS gas sensors are all adjacent MEMS gas sensors of the first judgment criterion or the second judgment criterion, there are two hydrogen leakage areas; and so on, the number of hydrogen leakage areas is obtained;

[0044] When there are S hydrogen leakage areas, S is an integer greater than or equal to 2, and starting from the first benchmark, its three adjacent MEMS gas sensors are matched in sequence; then the three adjacent MEMS gas sensors of the second benchmark are matched in sequence; and so on, until the three adjacent MEMS gas sensors of the Sth benchmark are matched.

[0045] In the detection method for indoor hydrogen leakage positioning,

[0046] Step 2, using the EEMD-GRNN ensemble empirical mode decomposition-generalized regression neural network model to process the predicted hydrogen concentration response curve over time, and obtain a smooth fitting curve with white noise removed;

[0047] Step 3: Combine the Gaussian distribution solution with Fick’s second law , perform nonlinear fitting to obtain the best fitting parameters , D and r, and calculate the distance r between the MEMS gas sensor and the hydrogen leakage location;

[0048] Step 4, repeat steps 2 and 3 to obtain the predicted distance r between the MEMS gas sensor corresponding to each hydrogen leakage area and the hydrogen leakage position, and obtain the three-dimensional coordinates of the MEMS gas sensor; obtain the coordinates of the location where the hydrogen leakage occurs according to the three-sided positioning method.

[0049] A MEMS gas sensor detection system for indoor hydrogen leak location includes:

[0050] A baseline compensation module, which uses an EEMD-GRNN ensemble empirical mode decomposition-generalized regression neural network model to obtain the drift of the real-time response baseline of the MEMS gas sensor;

[0051] A detection and analysis module uses a MEMS gas sensor to detect hydrogen of unknown concentration and obtain a response signal; uses q centralized processing units to collect the response signals obtained by all MEMS gas sensors, where q is an integer greater than or equal to 2, and transmits them to a computer; and obtains a curve of hydrogen concentration changing over time after processing the response signal;

[0052] A region identification module, which identifies the number of hydrogen leakage regions according to the curve of change of hydrogen concentration over time;

[0053] The leakage location module calculates the distance r between the MEMS gas sensor and the hydrogen leakage location based on the Gaussian distribution solution of Fick's second law equation, and obtains the predicted location of the indoor hydrogen leakage using the three-sided positioning method.

[0054] Compared with the prior art, the present invention has the following advantages: the present invention realizes quantitative measurement of indoor hydrogen leakage through MEMS gas sensors, and realizes accurate positioning of indoor hydrogen leakage position by building a three-dimensional network of MEMS gas sensors combined with trilateral positioning method. According to the location information of hydrogen leakage, fault information can be checked in time, and maintenance work can be performed on hydrogen storage equipment to ensure safe and stable operation of equipment and reduce risks. The detection method and system for indoor hydrogen leakage positioning using MEMS gas sensors provided by the present invention can be applied to hydrogen tank storage rooms, hydrogen transportation pipelines or indoor spaces with hydrogen storage equipment. The real-time response baseline of the MEMS gas sensor is processed by the EEMD-GRNN set empirical mode decomposition-generalized regression neural network model, and baseline compensation is realized by difference, thereby solving the interference of environmental factors on the detection of target gas concentration by the MEMS gas sensor, and multiple hydrogen leakage positions can be located at the same time by identifying the number of hydrogen leakage areas, and the indoor hydrogen concentration is detected in real time by the MEMS gas sensor, thereby realizing the real-time positioning of the hydrogen leakage position. The technician can obtain the occurrence locations of multiple hydrogen leaks at the same time, thereby saving the time and energy of the technician to identify fault information and improving the maintenance efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] By reading the detailed description of the preferred specific embodiments below, various other advantages and benefits of the present invention will become clear to those of ordinary skill in the art. The drawings in the specification are only for the purpose of illustrating the preferred embodiments and are not considered to be limitations of the present invention. Obviously, the drawings described below are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can also be obtained based on these drawings without creative work. Moreover, the same reference numerals are used to represent the same components throughout the drawings.

[0056] In the attached picture:

[0057] Figure 1 It is a schematic diagram of the MEMS gas sensor detection method for indoor hydrogen leakage positioning in the present invention;

[0058] Figure 2 It is a schematic diagram of a MEMS gas sensor detection system for indoor hydrogen leakage positioning in the present invention;

[0059] Figure 3 This is a schematic diagram of the process of using a MEMS gas sensor to locate a single hydrogen leakage area in a room in the first embodiment of the present invention;

[0060] Figure 4This is a schematic diagram of the process of using a MEMS gas sensor to locate when there are multiple hydrogen leakage occurrence areas indoors in the second embodiment of the present invention.

[0061] The present invention is further explained below in conjunction with the accompanying drawings and embodiments. DETAILED DESCRIPTION

[0062] The specific embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although the specific embodiments of the present invention are shown in the accompanying drawings, it should be understood that the present invention can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided in order to enable a more thorough understanding of the present invention and to fully convey the scope of the present invention to those skilled in the art.

[0063] It should be noted that certain words are used in the specification and claims to refer to specific components. Those skilled in the art should understand that technicians may use different nouns to refer to the same component. This specification and claims do not use the difference in nouns as a way to distinguish components, but use the difference in the functions of the components as the criterion for distinction. As mentioned throughout the specification and claims, "including" or "comprising" is an open term, so it should be interpreted as "including but not limited to". The subsequent description of the specification is a preferred embodiment of the present invention, but the description is based on the general principles of the specification and is not intended to limit the scope of the present invention. The scope of protection of the present invention shall be determined by the attached claims.

[0064] To facilitate understanding of the embodiments of the present invention, further explanation will be given below by taking specific embodiments as examples in conjunction with the accompanying drawings, and each of the accompanying drawings does not constitute a limitation on the embodiments of the present invention.

[0065] like Figures 1 to 4 As shown, the detection method for indoor hydrogen leakage location includes the following steps:

[0066] Step 1: Build a three-dimensional network of MEMS gas sensors indoors and obtain the coordinates (x, y, z) of each node where the MEMS gas sensor is located;

[0067] Step 2: Use MEMS gas sensors to measure the response characteristics of indoor gas, use EEMD-GRNN ensemble empirical mode decomposition-generalized regression neural network model to perform baseline compensation, and obtain the curve of hydrogen concentration change over time of each MEMS gas sensor;

[0068] Step 3: Use the Gaussian distribution solution of Fick's second law equation to fit the curve of the change of hydrogen concentration over time to obtain the best fitting parameters, so as to obtain the distance between the MEMS gas sensor and the hydrogen leakage position;

[0069] Step 4: Obtain the location of the indoor hydrogen leakage based on the distance between the MEMS gas sensor and the hydrogen leakage location.

[0070] In a preferred implementation manner of the detection method for indoor hydrogen leakage locating, in step 1, the three-dimensional network of the MEMS gas sensor is composed of n identical MEMS gas sensors placed at different nodes indoors, n is an integer greater than or equal to 4, and a three-dimensional coordinate is established with the MEMS gas sensor on one of the nodes as the origin, and the coordinates (x, y, z) of the MEMS gas sensors at all other nodes are obtained.

[0071] In a preferred embodiment of the detection method for indoor hydrogen leakage location, step 2 comprises:

[0072] Step 2.1, the MEMS gas sensor is selected as a sensor for detecting hydrogen as the target gas;

[0073] Step 2.2, calibrate the MEMS gas sensor, use the EEMD-GRNN ensemble empirical mode decomposition-generalized regression neural network model to denoise the baseline signal and then fit the baseline response signal, and compensate for the drift of the MEMS gas sensor response baseline by making a difference;

[0074] Step 2.21, calibrate the MEMS gas sensor, and use EEMD ensemble empirical decomposition to denoise the MEMS gas sensor response baseline signal;

[0075] Step 2.22, divide the denoised MEMS gas sensor response baseline signal into a training set and a validation set with a ratio of 4:1, and normalize the divided training data and validation data to reduce the amount of model calculation;

[0076] Step 2.23, set the radial basis function of the GRNN generalized regression neural network model to a Gaussian function, set the smoothing factor parameter, and use the training data set and the set smoothing factor parameter to build the GRNN generalized regression neural network model;

[0077] Step 2.24, use the GRNN generalized regression neural network model to predict the validation set data and calculate the mean square error between the predicted value and the true value;

[0078] Step 2.25, repeat steps 2.23-2.24, use different smoothing factor parameters, and calculate the mean square error between the predicted value and the true value;

[0079] Step 2.26, by comparing the mean square error between the predicted value and the true value obtained using different smoothing factor parameters, the optimal smoothing factor parameter is obtained, and a GRNN generalized regression neural network model is built to fit the denoised MEMS gas sensor response baseline signal;

[0080] Step 2.27, judging the fitting result by comparing the fitting curve with the original curve after denoising, increasing the smoothing factor when overfitting, and reducing the smoothing factor when underfitting; compensating for the drift of the MEMS gas sensor response baseline by making a difference;

[0081] The EEMD-GRNN ensemble empirical mode decomposition-generalized regression neural network model includes an input layer, a hidden layer, a summation layer, and an output layer; the input layer is responsible for receiving the baseline response signal of the MEMS gas sensor, the number of neurons in the hidden layer is equal to the number of samples in the training set, the summation layer includes a weighted summation layer and a nonlinear summation layer, and the output layer generates the final prediction result based on the result of the summation layer; the GRNN model usually does not require an iterative training process, and its optimization is mainly achieved by adjusting the smoothing factor of the neural network parameter. If the smoothing factor is too small, it will be overfitting, and if the smoothing factor is too large, it will be underfitting.

[0082] Step 2.3, use the MEMS gas sensor to obtain the response signal to the target gas, and after cleaning the data using the missing value processing algorithm and the outlier processing algorithm, obtain the curve of the change of hydrogen concentration over time.

[0083] In a preferred embodiment of the detection method for indoor hydrogen leakage location, in step 2.2,

[0084] Perform baseline calibration of MEMS gas sensor under air;

[0085] The response baseline change curve of the MEMS gas sensor over time is input into the EEMD-GRNN ensemble empirical mode decomposition-generalized regression neural network model, and the baseline signal is decomposed and recombined using EEMD to achieve the purpose of noise reduction, and the drift trend signal is obtained. The drift trend signal is modeled and fitted using the EEMD-GRNN ensemble empirical mode decomposition-generalized regression neural network model to obtain the prediction curve of the response baseline;

[0086] The predicted baseline is subtracted from the response curve of the MEMS gas sensor to hydrogen to complete the baseline compensation.

[0087] In a preferred embodiment of the detection method for indoor hydrogen leakage location, in step 2.3,

[0088] Use MEMS gas sensors to detect unknown concentrations of hydrogen and obtain response signals;

[0089] Install q centralized processing units indoors, where q is an integer greater than or equal to 2, and reasonably connect the MEMS gas sensors to the centralized processing units according to the locations of the MEMS gas sensors and the centralized processing units; the centralized processing units acquire response signals measured by each sensor and transmit them wirelessly to a computer;

[0090] After baseline compensation of the response signal of the MEMS gas sensor, the missing value processing algorithm and the outlier processing algorithm are used to clean the data. For the missing values ​​in the response signal, first identify the missing data points, which can be achieved through the ismissing function in MATLAB, and calculate the average of the two data points around it for interpolation; the median absolute deviation method is used to process the outliers in the target gas response signal, calculate the absolute deviation between each data point and the median, and find the median of these deviations, set the threshold to 3 times the median deviation, and remove the data points that exceed the threshold; obtain the curve of hydrogen concentration change over time.

[0091] In a preferred embodiment of the detection method for indoor hydrogen leakage location, step 3 comprises:

[0092] Step 3.1, the Gaussian distribution solution of Fick's second law equation is ,in is the concentration at a certain node in the room, is the hydrogen concentration at the point where the indoor hydrogen storage equipment leaks, r is the distance between the node and the hydrogen leakage location, D is the diffusion coefficient, and t is the diffusion time;

[0093] Step 3.2, using the Gaussian distribution solution to fit the curve of the hydrogen concentration changing with time, to obtain the best fitting parameters , D and r, thereby obtaining the distance r between the MEMS gas sensor and the indoor hydrogen leakage location;

[0094] Step 3.3, evaluate the fitting results, and evaluate the quality of the fitting results by calculating the residual, mean square error and root mean square error. Among them, the fitting accuracy can be improved by adjusting the model parameters of the EEMD-GRNN ensemble empirical mode decomposition-generalized regression neural network model or increasing the number of data points.

[0095] In a preferred embodiment of the detection method for indoor hydrogen leakage location, step 4 comprises:

[0096] Step 4.1, sort the MEMS gas sensors from small to large according to their distance r from the hydrogen leakage location;

[0097] Step 4.2, select four MEMS gas sensors that are closest to the hydrogen leak location and are not in the same plane as the positioning nodes of the three-side positioning method;

[0098] Step 4.3, draw spheres in ascending order with the positioning nodes as the sphere center and the distance r between them and the hydrogen leakage position as the radius. The first two spheres intersect to obtain a closed curve. The third sphere intersects with the closed curve obtained by the first two spheres to obtain two points. The point closest to the fourth sphere among these two points is the predicted position of the indoor hydrogen leakage position.

[0099] In a preferred embodiment of the detection method for indoor hydrogen leakage location, hydrogen leakage area identification includes:

[0100] According to the arrangement position of the MEMS gas sensors, the maximum distance Q between the three closest MEMS gas sensors and the n MEMS gas sensors is obtained. n , where n is an integer greater than or equal to 4, the largest Q n The value is recorded as M;

[0101] With the maximum Q n The value M is a judgment standard. If the distance between a certain MEMS gas sensor and another MEMS gas sensor is less than or equal to M, then the two MEMS gas sensors are adjacent MEMS gas sensors to each other.

[0102] The MEMS gas sensor that first generates a response signal is selected as a first judgment reference, and the MEMS gas sensors that subsequently generate response signals are judged in turn to be adjacent MEMS gas sensors of the first judgment reference;

[0103] If the subsequent three MEMS gas sensors are all adjacent MEMS gas sensors of the first judgment criterion, there is a single hydrogen leakage area; if one of the subsequent three MEMS gas sensors is not an adjacent MEMS gas sensor of the first judgment criterion, this MEMS gas sensor is used as the second judgment criterion to judge whether the subsequent MEMS gas sensors are adjacent MEMS gas sensors of the first judgment criterion or the second judgment criterion, until three adjacent MEMS gas sensors of the second judgment criterion appear, if the middle MEMS gas sensors are all adjacent MEMS gas sensors of the first judgment criterion or the second judgment criterion, there are two hydrogen leakage areas; and so on, the number of hydrogen leakage areas is obtained;

[0104] When there are S hydrogen leakage areas, S is an integer greater than or equal to 2, and starting from the first benchmark, its three adjacent MEMS gas sensors are matched in sequence; then the three adjacent MEMS gas sensors of the second benchmark are matched in sequence; and so on, until the three adjacent MEMS gas sensors of the Sth benchmark are matched.

[0105] In a preferred embodiment of the detection method for indoor hydrogen leakage location,

[0106] Step 2, using the EEMD-GRNN ensemble empirical mode decomposition-generalized regression neural network model to process the predicted hydrogen concentration response curve over time, and obtain a smooth fitting curve with white noise removed;

[0107] Step 3: Combine the Gaussian distribution solution with Fick’s second law , perform nonlinear fitting to obtain the best fitting parameters , D and r, and calculate the distance r between the MEMS gas sensor and the hydrogen leakage location;

[0108] Step 4, repeat steps 2 and 3 to obtain the predicted distance r between the MEMS gas sensor corresponding to each hydrogen leakage area and the hydrogen leakage position, and obtain the three-dimensional coordinates of the MEMS gas sensor; obtain the coordinates of the location where the hydrogen leakage occurs according to the three-sided positioning method.

[0109] A MEMS gas sensor detection system for indoor hydrogen leak location includes:

[0110] A baseline compensation module, which uses an EEMD-GRNN ensemble empirical mode decomposition-generalized regression neural network model to obtain the drift of the real-time response baseline of the MEMS gas sensor;

[0111] A detection and analysis module uses a MEMS gas sensor to detect hydrogen of unknown concentration and obtain a response signal; uses q centralized processing units to collect the response signals obtained by all MEMS gas sensors, where q is an integer greater than or equal to 2, and transmits them to a computer; and obtains a curve of hydrogen concentration changing over time after processing the response signal;

[0112] A region identification module, which identifies the number of hydrogen leakage regions according to the curve of change of hydrogen concentration over time;

[0113] The leakage location module calculates the distance r between the MEMS gas sensor and the hydrogen leakage location based on the Gaussian distribution solution of Fick's second law equation, and obtains the predicted location of the indoor hydrogen leakage using the three-sided positioning method.

[0114] Embodiment 1

[0115] like Figure 3As shown, it is a schematic diagram of the process of using a three-dimensional network of MEMS gas sensors to locate a single hydrogen leak area indoors. The process includes:

[0116] The concentration of hydrogen is detected using a MEMS gas sensor to obtain a hydrogen response signal, and then baseline compensation is performed. After the data is processed using a data cleaning algorithm, a curve of the relationship between hydrogen concentration and time is obtained;

[0117] The baseline compensation process includes:

[0118] Perform baseline calibration of MEMS gas sensor under air;

[0119] The response baseline change curve of the MEMS gas sensor over time is input into the EEMD-GRNN ensemble empirical mode decomposition-generalized regression neural network model. The baseline signal is decomposed and recombined using EEMD to achieve the purpose of noise reduction, and the drift trend signal is obtained. The drift trend signal is modeled and fitted using GRNN to obtain the prediction curve of the response baseline.

[0120] Subtract the predicted baseline from the response curve of the MEMS gas sensor to hydrogen to complete the baseline compensation;

[0121] Data cleaning algorithms include missing value processing algorithms and outlier processing algorithms.

[0122] The number of hydrogen leakage areas is determined based on the relationship curve between hydrogen concentration and time, which includes:

[0123] According to the arrangement position of the MEMS gas sensors, the maximum distance Q between the three closest MEMS gas sensors and the n MEMS gas sensors is obtained. n , where n is an integer greater than or equal to 4, the largest Q n The value is recorded as M;

[0124] With the maximum Q n The value M is a judgment standard. If the distance between a certain MEMS gas sensor and another MEMS gas sensor is less than or equal to M, then the two MEMS gas sensors are adjacent MEMS gas sensors to each other.

[0125] The MEMS gas sensor that first generates a response signal is selected as a first judgment reference, and the MEMS gas sensors that subsequently generate response signals are judged in turn to be adjacent MEMS gas sensors of the first judgment reference;

[0126] If the subsequent three MEMS gas sensors are all adjacent MEMS gas sensors of the first judgment criterion, there is a single hydrogen leakage area; if there is an adjacent MEMS gas sensor that is not the first judgment criterion among the subsequent three MEMS gas sensors, this MEMS gas sensor is used as the second judgment criterion to judge whether the subsequent MEMS gas sensors are adjacent MEMS gas sensors of the first judgment criterion or the second judgment criterion, until three adjacent MEMS gas sensors of the second judgment criterion appear. If the middle MEMS gas sensors are all adjacent MEMS gas sensors of the first judgment criterion or the second judgment criterion, there are two hydrogen leakage areas; and so on, the specific number of hydrogen leakage areas is obtained.

[0127] When the hydrogen leakage occurs in a single area, based on the relationship curve between hydrogen concentration and time, combined with the Gaussian distribution solution of Fick's second law equation, the predicted distance r between the MEMS gas sensor and the hydrogen leakage location is obtained, which includes:

[0128] Step 1: Use the EEMD-GRNN ensemble empirical mode decomposition-generalized regression neural network model to process the predicted response curve of the target gas concentration over time to obtain a smooth fitting curve with white noise removed;

[0129] Step 2: Combine the Gaussian distribution solution with Fick’s second law , perform nonlinear fitting to obtain the best fitting parameters, that is, , D and r, and calculate the distance r between the MEMS gas sensor and the hydrogen leakage location;

[0130] Step 3: Repeat steps 1 and 2 to obtain the predicted distance r between the MEMS gas sensor corresponding to each hydrogen leakage area and the hydrogen leakage position, and obtain the precise three-dimensional coordinates of the MEMS gas sensor.

[0131] Based on the predicted distance r between the MEMS gas sensor and the hydrogen leak location, the predicted location of the hydrogen leak is obtained using the trilateral positioning method, which includes:

[0132] Step 1: sort the MEMS gas sensors from small to large according to the distance r between them and the hydrogen leakage location;

[0133] Step 2: Select four MEMS gas sensors that are closest to the hydrogen leak location and are not in the same plane as the positioning nodes of the three-side positioning method;

[0134] Step three, use these nodes as the sphere center and the distance r from them to the hydrogen leakage location as the radius to draw spheres in ascending order. The intersection of the first two spheres can obtain a closed curve. The intersection of the third sphere and the closed curve obtained by the first two spheres can obtain two points. The point closest to the fourth sphere among these two points is the predicted position of the indoor hydrogen leakage location.

[0135] Embodiment 2

[0136] like Figure 4 As shown, it is a schematic diagram of the process of using a three-dimensional network of MEMS gas sensors to locate when there are multiple hydrogen leakage areas indoors. The process includes:

[0137] The concentration of hydrogen is detected using a MEMS gas sensor to obtain a hydrogen response signal, and then baseline compensation is performed. After the data is processed using a data cleaning algorithm, a curve of the relationship between hydrogen concentration and time is obtained;

[0138] The baseline compensation process includes:

[0139] Perform baseline calibration of MEMS gas sensor under air;

[0140] The response baseline change curve of the MEMS gas sensor over time is input into the EEMD-GRNN ensemble empirical mode decomposition-generalized regression neural network model. The baseline signal is decomposed and recombined using EEMD to achieve the purpose of noise reduction, and the drift trend signal is obtained. The drift trend signal is modeled and fitted using GRNN to obtain the prediction curve of the response baseline.

[0141] Subtract the predicted baseline from the response curve of the MEMS gas sensor to hydrogen to complete the baseline compensation;

[0142] The data cleaning algorithm includes a missing value processing algorithm and an outlier processing algorithm.

[0143] The number of hydrogen leakage areas is determined based on the relationship curve between hydrogen concentration and time. The process includes:

[0144] According to the arrangement position of the MEMS gas sensors, the maximum distance Q between the three closest MEMS gas sensors and the n MEMS gas sensors is obtained. n , where n is an integer greater than or equal to 4, the largest Q n The value is recorded as M;

[0145] With the maximum Q nThe value M is a judgment standard. If the distance between a certain MEMS gas sensor and another MEMS gas sensor is less than or equal to M, then the two MEMS gas sensors are adjacent MEMS gas sensors to each other.

[0146] The MEMS gas sensor that first generates a response signal is selected as a first judgment reference, and the MEMS gas sensors that subsequently generate response signals are judged in turn to be adjacent MEMS gas sensors of the first judgment reference;

[0147] If the subsequent three MEMS gas sensors are all adjacent MEMS gas sensors of the first judgment criterion, there is a single hydrogen leakage area; if there is an adjacent MEMS gas sensor that is not the first judgment criterion among the subsequent three MEMS gas sensors, this MEMS gas sensor is used as the second judgment criterion to judge whether the subsequent MEMS gas sensors are adjacent MEMS gas sensors of the first judgment criterion or the second judgment criterion, until three adjacent MEMS gas sensors of the second judgment criterion appear. If the middle MEMS gas sensors are all adjacent MEMS gas sensors of the first judgment criterion or the second judgment criterion, there are two hydrogen leakage areas; and so on, the specific number of hydrogen leakage areas is obtained.

[0148] When there are multiple areas where hydrogen leakage occurs, the MEMS gas sensors corresponding to different hydrogen leakage areas are matched. The specific process includes: when there are S hydrogen leakage areas, S is an integer greater than or equal to 2, and starting from the first benchmark, its three adjacent MEMS gas sensors are matched in sequence; then the three adjacent MEMS gas sensors of the second benchmark are matched in sequence; and so on, until the three adjacent MEMS gas sensors of the Sth benchmark are matched.

[0149] Based on the relationship curve between hydrogen concentration and time, combined with the Gaussian distribution solution of Fick's second law equation, the predicted distance r between the MEMS gas sensor and the hydrogen leakage position corresponding to each hydrogen leakage area is obtained in turn. The process includes:

[0150] Step 1: Use the EEMD-GRNN ensemble empirical mode decomposition-generalized regression neural network model to process the time response curve of the target gas concentration predicted by the MEMS gas sensor to obtain a smooth fitting curve with white noise removed;

[0151] Step 2: Combine the Gaussian distribution solution with Fick’s second law , perform nonlinear fitting to obtain the best fitting parameters, that is, , D and r, and calculate the distance r between the MEMS gas sensor and the hydrogen leakage location;

[0152] Step 3: Repeat steps 1 and 2 to obtain the predicted distance r between the MEMS gas sensor corresponding to each hydrogen leakage area and the hydrogen leakage position, and obtain the precise three-dimensional coordinates of the MEMS gas sensor.

[0153] Based on the predicted distance r between the MEMS gas sensor corresponding to each hydrogen leakage area and the hydrogen leakage position, the predicted position of each hydrogen leakage position is obtained by using the trilateral positioning method. The process includes:

[0154] Step 1: sort the MEMS gas sensors corresponding to each hydrogen leakage area from small to large according to the distance r between the MEMS gas sensors and the hydrogen leakage location;

[0155] Step 2: Select four MEMS gas sensors that are closest to the hydrogen leak location and are not in the same plane as the positioning nodes of the three-side positioning method;

[0156] Step 3: Use these nodes as the sphere center and the distance r from the hydrogen leakage location as the radius to draw spheres in ascending order. The intersection of the first two spheres can obtain a closed curve. The intersection of the third sphere and the closed curve obtained by the first two spheres can obtain two points. The point closest to the fourth sphere of these two points is the predicted position of the indoor hydrogen leakage location.

[0157] Step 4: Repeat steps 2 and 3 to obtain the predicted position of the hydrogen leakage position corresponding to each hydrogen leakage area.

[0158] The present invention realizes quantitative measurement of indoor hydrogen leakage by combining MEMS gas sensors with artificial intelligence algorithms, and realizes accurate positioning of indoor hydrogen leakage positions by building a three-dimensional network of MEMS gas sensors combined with a trilateral positioning method. According to the location information of hydrogen leakage, technicians can timely check fault information, perform maintenance work on hydrogen storage equipment, ensure safe and stable operation of equipment, and reduce risks. The detection method and system for indoor hydrogen leakage positioning using MEMS gas sensors provided by the present invention can be applied to hydrogen tank storage rooms, hydrogen transportation pipelines, or indoor spaces with hydrogen storage equipment. The real-time response baseline of the MEMS gas sensor is processed by the EEMD-GRNN set empirical mode decomposition-generalized regression neural network model, and baseline compensation is realized by difference, thereby solving the interference of environmental factors on the detection of target gas concentration by the MEMS gas sensor, and multiple hydrogen leakage positions can be located at the same time by identifying the number of hydrogen leakage areas, and the indoor hydrogen concentration is detected in real time by the MEMS gas sensor, thereby realizing the real-time positioning of the hydrogen leakage position. Technicians can obtain the occurrence locations of multiple hydrogen leaks at the same time, thereby saving the time and energy of technicians in identifying fault information and improving maintenance efficiency.

[0159] Although the embodiments of the present invention are described above in conjunction with the accompanying drawings, the present invention is not limited to the above specific embodiments and application fields, and the above specific embodiments are only illustrative and instructive, rather than restrictive. A person of ordinary skill in the art can also make many forms under the guidance of this specification and without departing from the scope of protection of the claims of the present invention, all of which belong to the protection of the present invention.

Claims

1. A detection method for indoor hydrogen leakage location, characterized in that: The method comprises the following steps: Step 1: Build a three-dimensional network of MEMS gas sensors indoors and obtain the coordinates (x, y, z) of each node where the MEMS gas sensor is located; Step 2: Use MEMS gas sensors to measure the response characteristics of indoor gas, use EEMD-GRNN ensemble empirical mode decomposition-generalized regression neural network model to perform baseline compensation, and obtain the curve of hydrogen concentration change over time of each MEMS gas sensor; Step 3: Use the Gaussian distribution solution of Fick's second law equation to fit the curve of the change of hydrogen concentration over time to obtain the best fitting parameters, so as to obtain the distance between the MEMS gas sensor and the hydrogen leakage position; Step 4: Obtain the location of the indoor hydrogen leakage based on the distance between the MEMS gas sensor and the hydrogen leakage location.

2. The detection method for indoor hydrogen leakage positioning according to claim 1, characterized in that: Preferably, in step 1, the three-dimensional network of the MEMS gas sensor is composed of n identical MEMS gas sensors placed at different nodes in the room, where n is an integer greater than or equal to 4, and a three-dimensional coordinate is established with the MEMS gas sensor on one of the nodes as the origin, and the coordinates (x, y, z) of the MEMS gas sensors at all other nodes are obtained.

3. The detection method for indoor hydrogen leakage positioning according to claim 1, characterized in that: Step 2 includes, Step 2.1, the MEMS gas sensor is selected as a sensor for detecting hydrogen as the target gas; Step 2.2, calibrate the MEMS gas sensor, use the EEMD-GRNN ensemble empirical mode decomposition-generalized regression neural network model to denoise the baseline signal and then fit the baseline response signal, and compensate for the drift of the MEMS gas sensor response baseline by making a difference; Step 2.21, calibrate the MEMS gas sensor, and use EEMD ensemble empirical decomposition to denoise the MEMS gas sensor response baseline signal; Step 2.22, divide the denoised MEMS gas sensor response baseline signal into a training set and a validation set with a ratio of 4:1, and normalize the divided training data and validation data to reduce the amount of model calculation; Step 2.23, set the radial basis function of the GRNN generalized regression neural network model to a Gaussian function, set the smoothing factor parameter, and use the training data set and the set smoothing factor parameter to build the GRNN generalized regression neural network model; Step 2.24, use the GRNN generalized regression neural network model to predict the validation set data and calculate the mean square error between the predicted value and the true value; Step 2.25, repeat steps 2.23-2.24, use different smoothing factor parameters, and calculate the mean square error between the predicted value and the true value; Step 2.26, by comparing the mean square error between the predicted value and the true value obtained using different smoothing factor parameters, the optimal smoothing factor parameter is obtained, and a GRNN generalized regression neural network model is built to fit the denoised MEMS gas sensor response baseline signal; Step 2.27, judging the fitting result by comparing the fitting curve with the original curve after denoising, increasing the smoothing factor when overfitting, and reducing the smoothing factor when underfitting; compensating for the drift of the MEMS gas sensor response baseline by making a difference; Step 2.3, use the MEMS gas sensor to obtain the response signal to the target gas, and after cleaning the data using the missing value processing algorithm and the outlier processing algorithm, obtain the curve of the change of hydrogen concentration over time.

4. The detection method for indoor hydrogen leakage positioning according to claim 1, characterized in that: In step 2.2, Perform baseline calibration of MEMS gas sensor under air; The response baseline change curve of the MEMS gas sensor over time is input into the EEMD-GRNN ensemble empirical mode decomposition-generalized regression neural network model, and the baseline signal is decomposed and recombined using EEMD to achieve the purpose of noise reduction, and the drift trend signal is obtained. The drift trend signal is modeled and fitted using the EEMD-GRNN ensemble empirical mode decomposition-generalized regression neural network model to obtain the prediction curve of the response baseline; The predicted baseline is subtracted from the response curve of the MEMS gas sensor to hydrogen to complete the baseline compensation.

5. The detection method for indoor hydrogen leakage positioning according to claim 1, characterized in that: In step 2.3, Use MEMS gas sensors to detect unknown concentrations of hydrogen and obtain response signals; Install q centralized processing units indoors, where q is an integer greater than or equal to 2, and reasonably connect the MEMS gas sensors to the centralized processing units according to the locations of the MEMS gas sensors and the centralized processing units; the centralized processing units acquire response signals measured by each sensor and transmit them wirelessly to a computer; After baseline compensation of the response signal of the MEMS gas sensor, the missing value processing algorithm and the outlier processing algorithm are used to clean the data to obtain the curve of hydrogen concentration changing with time.

6. The detection method for indoor hydrogen leakage location according to claim 1, characterized in that: Step 3 includes, Step 3.1, the Gaussian distribution solution of Fick's second law equation is ,in is the concentration at a certain node in the room, is the hydrogen concentration at the point where the indoor hydrogen storage equipment leaks, r is the distance between the node and the hydrogen leakage location, D is the diffusion coefficient, and t is the diffusion time; Step 3.2, using the Gaussian distribution solution to fit the curve of the hydrogen concentration changing with time, to obtain the best fitting parameters , D and r, thereby obtaining the distance r between the MEMS gas sensor and the indoor hydrogen leakage location; Step 3.3, evaluate the fitting results, and evaluate the quality of the fitting results by calculating the residual, mean square error and root mean square error. Among them, the fitting accuracy can be improved by adjusting the model parameters of the EEMD-GRNN ensemble empirical mode decomposition-generalized regression neural network model or increasing the number of data points.

7. The detection method for indoor hydrogen leakage location according to claim 1, characterized in that: Step 4 includes, Step 4.1, sort the MEMS gas sensors from small to large according to their distance r from the hydrogen leakage location; Step 4.2, select four MEMS gas sensors that are closest to the hydrogen leak location and are not in the same plane as the positioning nodes of the three-side positioning method; Step 4.3, draw spheres in ascending order with the positioning nodes as the sphere center and the distance r between them and the hydrogen leakage position as the radius. The first two spheres intersect to obtain a closed curve. The third sphere intersects with the closed curve obtained by the first two spheres to obtain two points. The point closest to the fourth sphere among these two points is the predicted position of the indoor hydrogen leakage position.

8. The detection method for indoor hydrogen leakage location according to claim 1, characterized in that: Hydrogen leak area identification includes, According to the arrangement position of the MEMS gas sensors, the maximum distance Q between the three closest MEMS gas sensors and the n MEMS gas sensors is obtained. n , where n is an integer greater than or equal to 4, the largest Q n The value is recorded as M; With the maximum Q n The value M is a judgment standard. If the distance between a certain MEMS gas sensor and another MEMS gas sensor is less than or equal to M, then the two MEMS gas sensors are adjacent MEMS gas sensors to each other. The MEMS gas sensor that first generates a response signal is selected as a first judgment reference, and the MEMS gas sensors that subsequently generate response signals are judged in turn to be adjacent MEMS gas sensors of the first judgment reference; If the subsequent three MEMS gas sensors are all adjacent MEMS gas sensors of the first judgment criterion, there is a single hydrogen leakage area; if one of the subsequent three MEMS gas sensors is not an adjacent MEMS gas sensor of the first judgment criterion, this MEMS gas sensor is used as the second judgment criterion to judge whether the subsequent MEMS gas sensors are adjacent MEMS gas sensors of the first judgment criterion or the second judgment criterion, until three adjacent MEMS gas sensors of the second judgment criterion appear. If the middle MEMS gas sensors are all adjacent MEMS gas sensors of the first judgment criterion or the second judgment criterion, there are two hydrogen leakage areas; And so on, the number of hydrogen leakage areas is obtained; When there are S hydrogen leakage areas, S is an integer greater than or equal to 2, and starting from the first benchmark, its three adjacent MEMS gas sensors are matched in sequence; then the three adjacent MEMS gas sensors of the second benchmark are matched in sequence; and so on, until the three adjacent MEMS gas sensors of the Sth benchmark are matched.

9. The detection method for indoor hydrogen leakage location according to claim 8, characterized in that: Step 2, using the EEMD-GRNN ensemble empirical mode decomposition-generalized regression neural network model to process the predicted hydrogen concentration response curve over time, and obtain a smooth fitting curve with white noise removed; Step 3: Combine the Gaussian distribution solution with Fick’s second law , perform nonlinear fitting to obtain the best fitting parameters , D and r, and calculate the distance r between the MEMS gas sensor and the hydrogen leakage location; Step 4, repeat steps 2 and 3 to obtain the predicted distance r between the MEMS gas sensor corresponding to each hydrogen leakage area and the hydrogen leakage position, and obtain the three-dimensional coordinates of the MEMS gas sensor; obtain the coordinates of the location where the hydrogen leakage occurs according to the three-sided positioning method.

10. A MEMS gas sensor detection system for indoor hydrogen leak location, characterized in that: The system comprises, A baseline compensation module, which uses an EEMD-GRNN ensemble empirical mode decomposition-generalized regression neural network model to obtain the drift of the real-time response baseline of the MEMS gas sensor; A detection and analysis module uses a MEMS gas sensor to detect hydrogen of unknown concentration and obtain a response signal; uses q centralized processing units to collect the response signals obtained by all MEMS gas sensors, where q is an integer greater than or equal to 2, and transmits them to a computer; and obtains a curve of hydrogen concentration changing over time after processing the response signal; A region identification module, which identifies the number of hydrogen leakage regions according to the curve of change of hydrogen concentration over time; The leakage location module calculates the distance r between the MEMS gas sensor and the hydrogen leakage location based on the Gaussian distribution solution of Fick's second law equation, and obtains the predicted location of the indoor hydrogen leakage using the three-sided positioning method.

Images