Linkage alarm household laser methane detection alarm and detection system

Abstract

The invention discloses a linkage alarm household laser methane detection alarm and a detection system, relates to the technical field of gas leakage alarms, and solves the problems that a sensor in a combustible gas alarm is subjected to large cross interference of hydrocarbon combustibles and is poisoned when meeting silicon and sulfide, and the influence of vibration is large. The laser methane detection alarm comprises a light source emitter, an optical module, an illumination detector, a control module, a power supply module, an interface module, a shell and external equipment, the laser methane detection system comprises a plurality of laser methane detection alarms, a processing unit, an early warning unit and a mobile phone app display unit, gas leakage early warning is carried out on the methane concentration data and the light intensity data through a preset full-time prediction model, the alarm threshold values of the volumes of different monitoring areas are corrected, and then external equipment is adjusted to process a gas leakage area; the detection data of the sensor is processed through a specially designed data processing mode, so that the detection efficiency of the laser methane detection alarm is improved, and the problem of poor gas leakage prediction caused by cross interference and vibration influence of the sensor is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of gas leakage detection, and in particular to a linkage alarm household laser methane detection alarm and a detection system. Background Art

[0002] With the development of society and the improvement of people's living standards, the use of methane-containing fuels such as natural gas and liquefied petroleum gas in households is becoming more and more common. During the use of these fuels, if they are not properly managed or the equipment is aging, they may cause leakage, thereby causing safety accidents. In the past, due to the lack of effective monitoring methods, accidents such as poisoning, fire, and explosion caused by household gas leakage occurred from time to time. These accidents not only caused casualties and property losses, but also aroused widespread social concern about household gas safety.

[0003] Patent No. CN2013101660860 discloses an alarm intelligent linkage method, including the following steps: storing the position control information of the pan / tilt when the camera is aimed at the alarm point on the device management server; after the alarm server detects the alarm information of the alarm point, the alarm server sends the alarm information to the device management server and the storage server; after the device management server receives the alarm information, it rotates the pan / tilt to the position where the camera is aimed at the alarm point according to the position control information; after the storage server receives the alarm information, it stores the video. Through the linkage between the device management server and the pan / tilt, the problem of redundant camera installation or the camera not monitoring the alarm point in time when the alarm occurs is solved, while avoiding the problem of wasting video storage space and saving storage space.

[0004] Patent No. CN2020110334729 discloses an automatic fire alarm and linkage control system, including a main control board, an intelligent circuit board and a terminal device; the intelligent circuit board includes a communication module and a control module; the terminal device is used to obtain environmental information of the monitoring area and upload the environmental information to the communication module; the main control board is used to perform the main control task, and the main control task includes obtaining the fire situation in the monitoring area according to the environmental information transmitted by the communication module, and issuing corresponding action instructions according to the fire situation; the terminal device is also used to perform corresponding actions according to the action instructions transmitted by the communication module; the control module is used to detect whether the communication between the communication module and the main control board is normal. When the communication between the communication module and the main control board is detected to be abnormal, the control module is used to replace the main control board to perform the main control task; the control module is used to detect whether the communication between the communication module and the main control board is normal, and when it is abnormal, the control module replaces the main control board to perform the main control task, thereby improving reliability.

[0005] Although the combustible gas detection alarm in the above patent can detect, remind and link natural gas, the sensor in the combustible gas detection alarm is a semiconductor and catalytic element. Therefore, the sensor is subject to large cross-interference from hydrocarbon combustibles, and is greatly affected by silicon, sulfide poisoning and vibration, resulting in poor gas leak prediction. Summary of the invention

[0006] The purpose of the present invention is to provide a household laser methane detection alarm and detection system with a linkage alarm, which can use a preset full-time prediction model to predict the gas leak alarm threshold and early warning for the collected methane concentration data, and correct the alarm threshold of different monitoring area volumes to adjust external equipment to process the gas leak, and link the smart gas meter to share the alarm information; so as to improve the detection efficiency of the laser methane detection alarm, and make up for the problem that the existing sensor is poor in predicting gas leaks due to large cross-interference of hydrocarbon combustibles, silicon, sulfide poisoning and vibration.

[0007] The present invention utilizes the following technical solutions:

[0008] A linkage alarm household laser methane detection alarm, comprising a laser sensor, a control module, a power module, an interface module, a housing and external equipment; wherein:

[0009] Laser sensors are used to detect methane concentration data in different monitoring area volumes;

[0010] The control module is used to process the acquired methane concentration data and judge the methane concentration data with the alarm threshold; at the same time, according to the real-time correction of the alarm threshold of the volume of different monitoring areas, an alarm signal is generated based on the real-time methane concentration data and the corrected alarm threshold, and an alarm and ventilation discharge are performed through external equipment;

[0011] The power module is used to power the laser methane detection alarm and manage the battery charging and discharging status;

[0012] Interface module, used to transmit methane concentration data and alarm thresholds to external devices for display and setting;

[0013] The housing is used to protect the internal components of the laser methane detection alarm; the internal components include a laser sensor, a control module, a power module and an interface module;

[0014] External equipment, used for alarm and ventilation discharge according to the alarm signal sent by the control module;

[0015] The power module is connected to the laser sensor, the control module and the interface module respectively through the power management circuit; the laser sensor is connected to the control module, and the control module is connected to the external device through the interface module. All internal components are installed in the shell.

[0016] Preferably, the external device includes a buzzer, a display device, a gas valve and an exhaust control fan.

[0017] A household laser methane detection system with linkage alarm includes a processing unit, an early warning unit, an App display unit, several laser methane detection alarms and several smart gas meters; wherein,

[0018] Laser methane detection alarm, used to obtain the monitored methane concentration data in the monitoring area;

[0019] Smart gas meters, which measure gas usage in homes and share alarm information from laser methane detectors;

[0020] The processing unit is used to pre-process the methane concentration data and gas usage, and form full-time data blocks, holiday data blocks and working time data after division, classification, cleaning, association, extraction and compression;

[0021] The early warning unit is used to predict the gas leakage alarm threshold value for the pre-processed methane concentration data and gas usage according to the volume of the monitoring area and the gas usage habits, and to correct the corresponding laser methane detection alarm threshold value according to the gas leakage alarm threshold value, and then send an alarm message to the smart gas meter when a gas leak occurs, so as to use the laser methane detection alarm to warn the monitoring area of ​​the gas leakage; the gas leakage prediction result includes the leakage amount and leakage time; the volume of the monitoring area is selected and determined by the assembler according to the installation space of each laser methane detection alarm;

[0022] The display unit is used to display the working status and methane concentration data of each laser methane detection alarm and external equipment.

[0023] Preferably, the laser methane detection alarm obtains the methane concentration data of the monitoring area through the laser sensor; the smart gas meter collects statistics on the gas usage of each user to form the gas usage habits of each user; the gas usage habits are the quarterly gas usage of the user in different gas usage periods.

[0024] Preferably, the process of preprocessing the methane concentration data and the gas usage by the processing unit is as follows:

[0025] S1: Divide the methane concentration data into several concentration data blocks according to the preset time interval; divide the gas usage into several usage data blocks according to the month;

[0026] S2: using a clustering algorithm to classify the internal data of the concentration data block and the usage data block according to the data type; the data type includes methane concentration and gas volume;

[0027] S3: Use data cleaning algorithms to clean and correct abnormal data in the classified internal data; data cleaning algorithms include data deduplication algorithms, data filling algorithms, data conversion algorithms, and data simplification algorithms;

[0028] S4: The internal data that has been cleaned is normalized by associating the methane concentration data with the gas usage according to the gas usage period to construct a full-time data block;

[0029] S5: extract and integrate all data of the corresponding time period in the full-time data block in time sequence according to holidays to construct holiday time data blocks and working time data blocks;

[0030] S6: using a data compression algorithm to losslessly compress the full-time data blocks, holiday period data blocks, and working period data blocks respectively.

[0031] Preferably, the early warning unit uses a preset full-time prediction model to predict the gas leak alarm threshold for the full-time data blocks, holiday period data blocks and working period data blocks formed by preprocessing, and corrects the alarm thresholds of different monitoring area volumes; the full-time prediction model includes a feature extraction layer, an iterative training layer and a recognition prediction layer; first, the feature extraction layer extracts features from the full-time data blocks, holiday period data blocks and working period data blocks respectively to form a full-time feature matrix, a holiday period feature matrix and a working period feature matrix; then, the iterative training layer performs several iterative training on the holiday period feature matrix and the working period feature matrix to form a work-holiday feature weight matrix; finally, the recognition prediction layer optimizes the work-holiday feature weight matrix according to the monitoring area volume and the full-time feature matrix, and corrects the alarm thresholds of different monitoring area volumes.

[0032] Preferably, the feature extraction layer uses three extraction branches to perform feature extraction on the all-time data blocks, holiday period data blocks and working period data blocks respectively to form the all-time feature matrix, the holiday period feature matrix and the working period feature matrix; the first extraction branch uses 1 3×3 convolution layer, 1 5×5 convolution layer, 1 convolution block composed of 3×1 convolution layer and 1×3 convolution layer and 1 3×3 average pooling layer to extract features from the all-time data blocks, and then obtain the all-time feature matrix; the second extraction branch uses 2 3×3 convolution layers, 1 5×5 convolution layer, 1 convolution block composed of 3×1 convolution layer and 1×3 convolution layer and 1 5×5 maximum pooling layer to extract features from the holiday period data blocks, and then obtain the holiday period feature matrix; the third extraction branch uses 2 3×3 convolution layers, 1 5×5 convolution layer and 2 series 5×5 maximum pooling layers to extract features from the working period data blocks, and then obtain the working period feature matrix.

[0033] Preferably, the iterative training layer uses 2 serial 3×3 convolutional layers, 2 5×5 adaptive pooling layers, 2 InceptionV2 blocks, 2 InceptionV4 blocks, 4 3×3 maximum pooling layers and 1 ReLU activation function to perform Hadamard operations on the holiday period feature matrix and the work period feature matrix, and then performs several learning and training to obtain the work-holiday feature weight matrix.

[0034] Preferably, the recognition prediction layer optimizes the gas leak prediction results of the methane-illumination feature weight matrix using 2 shuffle blocks, 1 1×1 convolutional layer, 2 fully connected layers, 2 5×5 average pooling layers and a feedback optimization function, and then verifies the work-holiday feature weight matrix through the full-time feature matrix, and corrects the alarm thresholds of different monitoring area volumes according to the verification results.

[0035] The present invention predicts and warns gas leak alarm thresholds for lossless data blocks formed by preprocessing through a preset full-time prediction model, and corrects the alarm thresholds of different monitoring area volumes to adjust external equipment to process gas leaks; and links smart gas meters to share alarm information; to improve the detection efficiency of laser methane detection alarms, and to compensate for the poor gas leak prediction problems of existing sensors due to large cross-interference from hydrocarbon combustibles, silicon, sulfide poisoning and vibration. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or related technologies, the drawings required for use in the embodiments or related technical descriptions are briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0037] Figure 1 This is the principle block diagram of the household laser methane detection system with linkage alarm. DETAILED DESCRIPTION

[0038] The present invention is described in detail below with reference to the accompanying drawings and embodiments:

[0039] like Figure 1 As shown, the invention discloses a linkage alarm household laser methane detection alarm, a laser sensor, a control module, a power module, an interface module, a housing and external equipment; wherein,

[0040] Laser sensors are used to detect methane concentration data in different monitoring area volumes;

[0041] The control module is used to process the acquired methane concentration data and judge the methane concentration data with the alarm threshold; at the same time, according to the real-time correction of the alarm threshold of the volume of different monitoring areas, an alarm signal is generated based on the real-time methane concentration data and the corrected alarm threshold, and an alarm and ventilation discharge are performed through external equipment;

[0042] The power module is used to power the laser methane detection alarm and manage the battery charging and discharging status;

[0043] Interface module, used to transmit methane concentration data and alarm thresholds to external devices for display and setting;

[0044] The housing is used to protect the internal components of the laser methane detection alarm; the internal components include a laser sensor, a control module, a power module and an interface module;

[0045] External equipment, used for alarm and ventilation discharge according to the alarm signal sent by the control module;

[0046] The power module is connected to the laser sensor, the control module and the interface module respectively through the power management circuit; the laser sensor is connected to the control module, the control module is connected to the external device through the interface module, and each internal component is installed in the housing;

[0047] In the present invention, the external equipment includes a buzzer, a display device, a gas valve and an exhaust control fan;

[0048] A household laser methane detection system with linkage alarm includes a processing unit, an early warning unit, an App display unit, several laser methane detection alarms and several smart gas meters; wherein,

[0049] Laser methane detection alarm, used to obtain the monitored methane concentration data in the monitoring area;

[0050] In the present invention, the laser methane detection alarm obtains the methane concentration data of the monitoring area through a laser sensor;

[0051] Smart gas meters, which measure gas usage in homes and share alarm information from laser methane detectors;

[0052] In the present invention, the smart gas meter collects statistics on the gas usage of each user to form the gas usage habits of each user; the gas usage habits are the quarterly gas usage of the user in different gas usage periods;

[0053] The processing unit is used to pre-process the methane concentration data and gas usage, and form full-time data blocks, holiday data blocks and working time data after division, classification, cleaning, association, extraction and compression;

[0054] In the present invention, the process of preprocessing the methane concentration data and the gas usage by the processing unit is as follows:

[0055] S1: Divide the methane concentration data into several concentration data blocks according to the preset time interval; divide the gas usage into several usage data blocks according to the month;

[0056] S2: using a clustering algorithm to classify the internal data of the concentration data block and the usage data block according to the data type; the data type includes methane concentration and gas volume;

[0057] S3: Use data cleaning algorithms to clean and correct abnormal data in the classified internal data; data cleaning algorithms include data deduplication algorithms, data filling algorithms, data conversion algorithms, and data simplification algorithms;

[0058] S4: The internal data that has been cleaned is normalized by associating the methane concentration data with the gas usage according to the gas usage period to construct a full-time data block;

[0059] S5: extract and integrate all data of the corresponding time period in the full-time data block in time sequence according to holidays to construct holiday time data blocks and working time data blocks;

[0060] S6: Losslessly compressing the full-time data blocks, holiday data blocks, and working time data blocks respectively using a data compression algorithm;

[0061] The early warning unit is used to predict the gas leakage alarm threshold value for the pre-processed methane concentration data and gas usage according to the volume of the monitoring area and the gas usage habits, and to correct the corresponding laser methane detection alarm threshold value according to the gas leakage alarm threshold value, and then send an alarm message to the smart gas meter when a gas leak occurs, so as to use the laser methane detection alarm to warn the monitoring area of ​​the gas leakage; the gas leakage prediction result includes the leakage amount and leakage time; the volume of the monitoring area is selected and determined by the assembler according to the installation space of each laser methane detection alarm;

[0062] The early warning unit uses the preset full-time prediction model to predict the gas leakage alarm threshold for the full-time data blocks, holiday data blocks and working time data blocks formed by preprocessing, and corrects the alarm thresholds of different monitoring area volumes; the full-time prediction model includes a feature extraction layer, an iterative training layer and an identification prediction layer;

[0063] In this embodiment, on weekdays, gas usage usually fluctuates regularly, such as peak gas usage in the morning and evening, which is related to residents' activities such as cooking and bathing; the average gas usage on weekdays may be relatively stable; when the gas usage is stable, smaller leaks may be easier to detect; the cause of the leak may be more related to daily use, such as equipment aging and improper operation;

[0064] Gas usage may be more irregular during holidays. As people spend more time at home, peak gas usage may last longer and fluctuate more. Average gas usage may be higher during holidays because family members are more concentrated at home. Due to the high gas usage, smaller leaks may be masked by normal gas usage and difficult to detect. Leaks may be more related to special activities, such as large-scale cooking and equipment overload.

[0065] Therefore, the full-time data block, holiday data block and working time data block are respectively extracted by different extraction branches, so as to obtain the full-time gas usage characteristics, holiday gas usage characteristics and working day gas usage characteristics;

[0066] First, the feature extraction layer extracts features from the full-time data blocks, holiday data blocks, and working time data blocks to form a full-time feature matrix, a holiday feature matrix, and a working time feature matrix;

[0067] In the present invention, the feature extraction layer uses three extraction branches to extract features from the full-time data blocks, holiday period data blocks and working period data blocks respectively, so as to form a full-time feature matrix, a holiday period feature matrix and a working period feature matrix;

[0068] In this embodiment, the first extraction branch uses one 3×3 convolutional layer, one 5×5 convolutional layer, one convolutional block consisting of a 3×1 convolutional layer and a 1×3 convolutional layer, and one 3×3 average pooling layer to extract features from the full-time data block, thereby obtaining a full-time feature matrix;

[0069]

[0070] Among them, Y 1 represents the full-time feature matrix, represents a 1×3 convolutional layer with a stride of 1, represents a 3×1 convolutional layer with a stride of 1, represents a 3×3 average pooling layer with a stride of 2, represents a 3×3 convolutional layer with a stride of 2, represents a 5×5 convolutional layer with a stride of 1, and T1 represents a full-time data block;

[0071] The second extraction branch uses two 3×3 convolutional layers, one 5×5 convolutional layer, one convolutional block consisting of a 3×1 convolutional layer and a 1×3 convolutional layer, and one 5×5 maximum pooling layer to extract features from the holiday period data block, and then obtain the holiday period feature matrix;

[0072]

[0073] Among them, Y 2 represents the holiday period feature matrix, represents a 1×3 convolutional layer with a stride of 2, represents a 3×1 convolutional layer with a stride of 2, represents a 5×5 maximum pooling layer with a stride of 2, represents a 3×3 convolutional layer with a stride of 1, represents a 3×3 convolutional layer with a stride of 2, represents element-wise multiplication, represents a 5×5 convolutional layer with a stride of 2, and T2 represents a holiday period data block;

[0074] The third extraction branch uses two 3×3 convolutional layers, one 5×5 convolutional layer, and two serially connected 5×5 maximum pooling layers to extract features from the working period data blocks, thereby obtaining a working period feature matrix;

[0075]

[0076] Among them, Y 3 represents the working period feature matrix, It indicates 2 5×5 maximum pooling layers connected in series with a stride of 2, and T3 indicates the working period data block;

[0077] Then, the iterative training layer performs several iterative trainings on the holiday period feature matrix and the working period feature matrix to form a working-holiday feature weight matrix;

[0078] In the present invention, the iterative training layer uses two serial 3×3 convolutional layers, two 5×5 adaptive pooling layers, one InceptionV2 block, one InceptionV4 block, two serial 5×5 convolutional layers and one ReLU activation function to perform Hadamard operations on the holiday period feature matrix and the work period feature matrix, and then performs several learning trainings to obtain the work-holiday feature weight matrix;

[0079]

[0080] Among them, Model 1 Represents the work-holiday feature weight matrix, AutoPool 5×5 2 () represents two 5×5 adaptive pooling layers in series, represents two 5×5 convolutional layers with a stride of 2 in series, represents two 3×3 convolutional layers with a stride of 2 in series;

[0081] Finally, the recognition prediction layer optimizes the work-holiday feature weight matrix according to the monitoring area volume and the full-time feature matrix, and modifies the alarm thresholds of different monitoring area volumes;

[0082] In the present invention, the recognition prediction layer optimizes the gas leakage prediction result of the methane-light feature weight matrix using 2 shuffle blocks, 2 1×1 convolutional layers, 2 fully connected layers, 2 5×5 average pooling layers and feedback optimization function, and then verifies the work-holiday feature weight matrix through the full-time feature matrix, and corrects the alarm threshold of different monitoring area volumes according to the verification results;

[0083]

[0084]

[0085] Among them, Model 2 Represents the recognition prediction layer, Loss represents the feedback optimization function, MeanPool 5×5 2 () represents two 5×5 average pooling layers in series, Shuff 2 Represents 2 shuffled blocks in series, represents a 1×1 convolutional layer with a step size of 2, V represents the volume of the monitoring area, σ represents the number of iterative training layers, G represents the quantization function, μ1 represents the calculation accuracy coefficient, and B σrepresents the iterative training loss function of each layer, μ2 represents the preset optimization weight coefficient, E σ Indicates the optimization efficiency of each layer’s iterative training;

[0086] A display unit is used to display the working status of each laser methane detection alarm and external equipment and methane concentration data;

[0087] A linkage alarm household laser methane detection alarm method comprises the following steps in sequence:

[0088] A: The power module supplies power to the laser methane detection alarm, and the laser methane detection alarm starts working;

[0089] B: The laser methane detection alarm detects the methane concentration data in the monitoring area, and the smart gas meters deployed in the monitoring area measure the gas usage in the home;

[0090] C: Transmitting the methane concentration data and gas usage to the processing unit of the laser methane detection system for pre-processing;

[0091] D: The pre-processed methane concentration data and gas usage are used to predict the gas leakage alarm threshold through the full-time prediction model preset in the early warning unit, and the alarm thresholds of different monitoring area volumes are corrected according to the gas leakage prediction results. At the same time, alarm information is sent to the smart gas meter according to the gas leakage prediction results, and the corrected alarm thresholds of different monitoring area volumes are transmitted to the laser methane detection alarm;

[0092] E: The control module of the laser methane detection alarm adjusts the working status of the external equipment according to the corrected alarm thresholds of the volumes of different monitoring areas;

[0093] F: The laser methane detection system displays the working status and methane concentration data of each laser methane detection alarm and external equipment on the display unit.

[0094] Example:

[0095] The power module supplies power to the laser methane detection alarm, and the laser methane detection alarm starts to work; the laser methane detection alarm detects the methane concentration data in the monitoring area, and the smart gas meter deployed in the monitoring area measures the gas usage in the household; at the same time, the methane concentration data and gas usage are transmitted to the laser methane detection system;

[0096] Then, the laser methane detection system receives the methane concentration data and the gas usage data, and uses the processing unit to divide the methane concentration data into several concentration data blocks according to the preset time interval; divides the gas usage into several usage data blocks according to the month; classifies the internal data of the data block according to the data type; uses the data cleaning algorithm to clean and correct the abnormal data of the classified internal data; the data cleaning algorithm includes a data deduplication algorithm, a data filling algorithm, a data conversion algorithm and a data simplification algorithm; the internal data that has completed data cleaning is associated and normalized with the methane concentration data and the gas usage according to the gas usage period to construct a full-time data block; all data of the corresponding time period in the full-time data block are extracted and integrated in time sequence according to holidays to construct holiday period data blocks and working period data blocks; the aggregated data block is losslessly compressed using a data compression algorithm to form a corresponding lossless data block; the lossless data block is transmitted to the early warning unit through wireless communication using a parallel heuristic transmission algorithm to predict the gas leak alarm threshold;

[0097] Subsequently, the early warning unit in the laser methane detection system uses a preset full-time prediction model to predict the gas leakage alarm threshold for the pre-processed methane concentration data and gas usage according to the volume of the monitoring area and gas usage habits, and corrects the alarm threshold according to the gas leakage prediction result, and then sends an alarm message to the smart gas meter, and issues an early warning for the monitoring area where each laser methane detection alarm is located; the full-time prediction model includes a feature extraction layer, an iterative training layer, and a recognition prediction layer;

[0098] First, the feature extraction layer uses three extraction branches to extract features from the full-time data blocks, holiday period data blocks and working period data blocks respectively to form the full-time feature matrix, holiday period feature matrix and working period feature matrix: the first extraction branch uses 1 3×3 convolution layer, 1 5×5 convolution layer, 1 convolution block composed of 3×1 convolution layer and 1×3 convolution layer and 1 3×3 average pooling layer to extract features from the full-time data blocks, and then obtain the full-time feature matrix; the second extraction branch uses 2 3×3 convolution layers, 1 5×5 convolution layer, 1 convolution block composed of 3×1 convolution layer and 1×3 convolution layer and 1 5×5 maximum pooling layer to extract features from the holiday period data blocks, and then obtain the holiday period feature matrix; the third extraction branch uses 2 3×3 convolution layers, 1 5×5 convolution layer and 2 series 5×5 maximum pooling layers to extract features from the working period data blocks. The feature extraction of the working period data block is performed to obtain the working period feature matrix; then, the iterative training layer uses 2 serial 3×3 convolutional layers, 2 5×5 adaptive pooling layers, 1 InceptionV2 block, 1 InceptionV4 block, 2 serial 5×5 convolutional layers and 1 ReLU activation function to perform Hadamard operation on the holiday period feature matrix and the working period feature matrix, and then performs several learning and training to obtain the working-holiday feature weight matrix; finally, the recognition prediction layer uses 2 shuffle blocks, 2 1×1 convolutional layers, 2 fully connected layers, 2 5×5 average pooling layers and feedback optimization function to optimize the gas leakage prediction result of the methane-light feature weight matrix, and then verifies the working-holiday feature weight matrix through the full-time feature matrix, and corrects the alarm threshold of different monitoring area volumes according to the verification results;

[0099] Finally, the control module in the laser methane detection alarm controls the working status of the laser methane detection alarm, the smart gas meter and external equipment according to the corrected alarm thresholds of the volumes of different monitoring areas; the display unit of the laser methane detection system displays the working status and methane concentration data of each laser methane detection alarm and external equipment.

Claims

1. A household laser methane detection alarm with linkage alarm, characterized by: It includes a laser sensor, a control module, a power module, an interface module, a housing and external equipment; wherein, Laser sensors are used to detect methane concentration data in different monitoring area volumes; The control module is used to process the acquired methane concentration data and judge the methane concentration data with the alarm threshold; at the same time, according to the real-time correction of the alarm threshold of the volume of different monitoring areas, an alarm signal is generated based on the real-time methane concentration data and the corrected alarm threshold, and an alarm and ventilation discharge are performed through external equipment; The power module is used to power the laser methane detection alarm and manage the battery charging and discharging status; Interface module, used to transmit methane concentration data and alarm thresholds to external devices for display and setting; The housing is used to protect the internal components of the laser methane detection alarm; the internal components include a laser sensor, a control module, a power module and an interface module; External equipment, used for alarm and ventilation discharge according to the alarm signal sent by the control module; The power module is connected to the laser sensor, the control module and the interface module respectively through the power management circuit; the laser sensor is connected to the control module, the control module is connected to the external device through the interface module, and each internal component is installed in the shell.

2. The linkage alarm household laser methane detection alarm according to claim 1, characterized in that: The external devices include a buzzer, a display screen, a gas valve and an exhaust control fan.

3. A linkage alarm household laser methane detection system, applied to the linkage alarm household laser methane detection alarm according to claims 1-2, characterized in that: It includes a processing unit, an early warning unit, a display unit, several laser methane detection alarms and several intelligent gas meters; among which, Laser methane detection alarm, used to obtain the monitored methane concentration data in the monitoring area; Smart gas meters, which measure gas usage in homes and share alarm information from laser methane detectors; The processing unit is used to pre-process the methane concentration data and gas usage, and form full-time data blocks, holiday data blocks and working time data after division, classification, cleaning, association, extraction and compression; The early warning unit is used to predict the gas leakage alarm threshold value for the pre-processed methane concentration data and gas usage according to the volume of the monitoring area and the gas usage habits, and to correct the corresponding laser methane detection alarm threshold value according to the gas leakage alarm threshold value, and then send an alarm message to the smart gas meter when a gas leak occurs, so as to use the laser methane detection alarm to warn the monitoring area of ​​the gas leakage; the gas leakage prediction result includes the leakage amount and leakage time; the volume of the monitoring area is selected and determined by the assembler according to the installation space of each laser methane detection alarm; The display unit is used to display the working status and methane concentration data of each laser methane detection alarm and external equipment.

4. The linkage alarm household laser methane detection system according to claim 3 is characterized by: The laser methane detection alarm obtains the methane concentration data of the monitoring area through the laser sensor; the smart gas meter counts the gas usage of each user to form the gas usage habits of each user; the gas usage habits are the quarterly gas usage of the user in different gas usage periods.

5. The linkage alarm household laser methane detection system according to claim 3 is characterized by: The process of preprocessing the methane concentration data and gas usage by the processing unit is as follows: S1: Divide the methane concentration data into several concentration data blocks according to the preset time interval; divide the gas usage into several usage data blocks according to the month; S2: using a clustering algorithm to classify the internal data of the concentration data block and the usage data block according to the data type; the data type includes methane concentration and gas volume; S3: Use data cleaning algorithms to clean and correct abnormal data in the classified internal data; data cleaning algorithms include data deduplication algorithms, data filling algorithms, data conversion algorithms, and data simplification algorithms; S4: The internal data that has been cleaned is normalized by associating the methane concentration data with the gas usage according to the gas usage period to construct a full-time data block; S5: extract and integrate all data of the corresponding time period in the full-time data block in time sequence according to holidays to construct holiday time data blocks and working time data blocks; S6: using a data compression algorithm to losslessly compress the full-time data blocks, holiday period data blocks, and working period data blocks respectively.

6. The linkage alarm household laser methane detection system according to claim 3 is characterized by: The early warning unit uses a preset full-time prediction model to predict the gas leakage alarm threshold for the full-time data block, holiday period data block and working period data block formed by preprocessing, and corrects the alarm threshold of different monitoring area volumes; The full-time prediction model includes a feature extraction layer, an iterative training layer, and an identification prediction layer; first, the feature extraction layer extracts features from the full-time data blocks, holiday data blocks, and working time data blocks respectively to form a full-time feature matrix, a holiday feature matrix, and a working time feature matrix; then, the iterative training layer performs several iterative trainings on the holiday feature matrix and the working time feature matrix to form a work-holiday feature weight matrix; Finally, the recognition and prediction layer optimizes the work-holiday feature weight matrix according to the monitoring area volume and the full-time feature matrix, and corrects the alarm thresholds of different monitoring area volumes.

7. The linkage alarm household laser methane detection system according to claim 6 is characterized by: The feature extraction layer uses three extraction branches to extract features from the full-time data blocks, holiday data blocks, and working time data blocks respectively, so as to form a full-time feature matrix, a holiday feature matrix, and a working time feature matrix; the first extraction branch uses a 3×3 convolution layer, a 5×5 convolution layer, a convolution block composed of a 3×1 convolution layer and a 1×3 convolution layer, and a 3×3 average pooling layer to extract features from the full-time data blocks, thereby obtaining a full-time feature matrix; The second extraction branch uses two 3×3 convolutional layers, one 5×5 convolutional layer, one convolutional block consisting of a 3×1 convolutional layer and a 1×3 convolutional layer, and one 5×5 maximum pooling layer to extract features from the holiday period data block, and then obtain the holiday period feature matrix; the third extraction branch uses two 3×3 convolutional layers, one 5×5 convolutional layer, and two series-connected 5×5 maximum pooling layers to extract features from the working period data block, and then obtain the working period feature matrix.

8. The linkage alarm household laser methane detection system according to claim 6 is characterized by: The iterative training layer uses 2 serial 3×3 convolutional layers, 2 5×5 adaptive pooling layers, 2 InceptionV2 blocks, 2 InceptionV4 blocks, 4 3×3 maximum pooling layers and 1 ReLU activation function to perform Hadamard operation on the holiday period feature matrix and the work period feature matrix, and then performs several learning trainings to obtain the work-holiday feature weight matrix.

9. The linkage alarm household laser methane detection system according to claim 6, characterized in that: The recognition prediction layer optimizes the gas leakage prediction results of the methane-illumination feature weight matrix using two shuffle blocks, one 1×1 convolutional layer, two fully connected layers, two 5×5 average pooling layers and a feedback optimization function, and then verifies the work-holiday feature weight matrix through the full-time feature matrix, and corrects the alarm thresholds of different monitoring area volumes according to the verification results.

Images