Intelligent gas valve based on narrowband Internet of Things communication mode

Abstract

The invention discloses an intelligent gas valve based on a narrowband Internet of Things communication mode, and relates to the technical field of intelligent gas valve control. The intelligent fuel gas valve based on the narrowband Internet of Things communication mode comprises a data acquisition and analysis unit used for acquiring a fuel gas conveying state and analyzing a fuel gas conveying abnormal response characteristic value, and a local abnormity judgment unit used for judging whether the fuel gas conveying state is higher than a preset fuel gas conveying abnormal response characteristic threshold value or not; the narrowband communication transmission unit is used for acquiring gas valve perimeter image data, and transmitting the gas valve perimeter image data based on a narrowband internet of things network in combination with gas transmission state data and corresponding characteristic values; the perimeter image analysis unit is used for receiving corresponding data and analyzing the pipe section conveying risk characteristic value based on a pre-trained perimeter state recognition model, the intelligent gas valve is intelligently controlled through the pipe section conveying risk characteristic value in the intelligent feedback control unit, and therefore distinguishing response to instantaneous fluctuation and persistent defects is achieved; and the control refinement is obviously improved.

Description

Technical Field

[0001] This invention relates to the field of gas intelligent valve control technology, specifically a gas intelligent valve based on narrowband Internet of Things communication. Background Technology

[0002] As the scale and complexity of gas pipeline networks continue to increase, valves, as key actuators for segmented isolation and emergency response, urgently need to possess online sensing, remote diagnostics, and automatic control capabilities. Current remote monitoring and control of gas pipeline network terminals typically consists of a sensing layer, an edge processing layer, and a communication access layer: the sensing layer includes sensors and valve opening and actuator status acquisition; the edge processing layer uses microcontrollers or embedded processors to perform data preprocessing and local logic control; the communication access layer often adopts cellular IoT access, among which narrowband IoT communication is characterized by wide coverage, low power consumption, low data rate, and massive connectivity, making it suitable for long-term online and low-energy operation of distributed terminals such as valves.

[0003] Existing technology, such as the patent application with publication number CN117128350A, discloses a gas self-closing valve control method, system, intelligent terminal, and storage medium, which includes: acquiring a gas usage signal and a gas pressure value in the gas pipeline; determining the gas usage time based on the gas usage signal; calculating the gas self-closing valve closing time based on the gas usage time and a preset gas usage duration; comparing and analyzing the pipeline gas pressure value with a preset normal gas pressure range to determine whether the pipeline gas pressure is within the normal gas pressure range; issuing a normal gas pressure prompt when the pipeline gas pressure value is within the preset normal gas pressure range; and instructing a preset control device to close the gas self-closing valve based on the normal gas pressure prompt instruction and the gas self-closing valve closing time. This application has the effect of improving the safety of gas pipelines.

[0004] Based on the above findings, the limitations of existing technologies include at least the following problems: Existing technologies struggle to effectively distinguish between transient fluctuations and persistent physical faults, which can easily lead to a decrease in control accuracy. Specifically, existing technologies do not introduce a chain of physical evidence related to the valve body perimeter, making it difficult to verify from the appearance and structural aspects whether the delivery anomaly is a transient fluctuation. Consequently, it is difficult to effectively distinguish between transient operating condition fluctuations and continuously evolving structural failures, which in turn can easily lead to false alarms in actual operation, such as misjudging transient disturbances as faults and triggering unnecessary shutdowns or flow restrictions, as well as missed alarms, such as being insensitive to early or boundary physical defects and failing to address them in a timely manner, causing gas supply interruptions and increased disturbances on the user side. This results in a coarse-grained triggering effect, making it difficult to meet the stability and controllability requirements under complex operating conditions. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a gas smart valve based on narrowband Internet of Things (IoT) communication, which solves the problem of existing technologies' inability to distinguish between transient and continuous states, resulting in coarse-grained valve control.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a smart gas valve based on narrowband IoT communication, comprising the following steps: a data acquisition and analysis unit, used to acquire real-time gas delivery status data of the smart gas valve and analyze the abnormal gas delivery response characteristic value of the smart gas valve; a local anomaly determination unit, used to determine whether the abnormal gas delivery response characteristic value of the smart gas valve is higher than a preset abnormal gas delivery response characteristic threshold; a narrowband communication transmission unit, used to acquire gas valve perimeter image data of the smart gas valve, combine the gas delivery status data and the corresponding abnormal gas delivery response characteristic value to generate a transmission data packet, and transmit it based on a narrowband IoT network; a perimeter image analysis unit, used to receive the gas valve perimeter image data and the abnormal gas delivery response characteristic value of the smart gas valve, and analyze the pipeline delivery risk characteristic value of the smart gas valve based on a pre-trained perimeter state recognition model; and an intelligent feedback control unit, used to intelligently control the smart gas valve based on the pipeline delivery risk characteristic value.

[0007] Furthermore, the gas delivery status data includes pipe pressure difference, pipe gas temperature, pipe turbulence vortex intensity, pipe water vapor content, pipe wall vibration amplitude, and flow velocity pulsation amplitude. The specific steps for analyzing the abnormal response characteristic values ​​of the set gas intelligent valve are as follows: Based on the gas delivery status data of the set gas intelligent valve, analyze the delivery characteristic set of the set gas intelligent valve, including gas flow stability characteristic values ​​and medium impact response characteristic values; based on the delivery characteristic set of the set gas intelligent valve, analyze the abnormal response characteristic values ​​of the set gas intelligent valve.

[0008] Furthermore, the specific steps for analyzing and setting the delivery characteristic set of the intelligent gas valve are as follows: Based on the differential pressure value, turbulent vortex intensity value, and velocity pulsation amplitude of the intelligent gas valve, analyze the stable gas flow characteristic value of the intelligent gas valve; based on the gas temperature value, water vapor content value, and pipe wall vibration amplitude of the intelligent gas valve, analyze the medium impact response characteristic value of the intelligent gas valve.

[0009] Furthermore, the specific steps for transmission based on the narrowband IoT network are as follows: establish a DTLS encrypted connection based on the NB-IoT modem; perform fragmentation processing on the transmission data packets based on the CoAP protocol to obtain several fragmented data packets; and transmit each fragmented data packet piece by piece based on the UDP / IP protocol stack.

[0010] Furthermore, the specific steps for analyzing and setting the pipeline transportation risk characteristic values ​​for the gas intelligent valve are as follows: Based on a pre-trained perimeter state recognition model and combined with the perimeter image data of the set gas smart valve, the perimeter disturbance degradation characteristic value of the set gas smart valve is analyzed; the abnormal gas delivery response characteristic value of the set gas smart valve is read, and combined with the perimeter disturbance degradation characteristic value, the pipeline delivery risk characteristic value of the set gas smart valve is analyzed.

[0011] Furthermore, the specific steps for analyzing the perimeter disturbance degradation characteristic values ​​of the set gas smart valve are as follows: input the gas valve perimeter image data of the set gas smart valve into the pre-trained perimeter state recognition model, analyze the perimeter abnormal response feature set of the set gas smart valve, including surface abnormal seepage feature values, perimeter seal fracture disturbance feature values, and structural deformation trace feature values; based on the perimeter abnormal response feature set of the set gas smart valve, analyze the perimeter disturbance degradation characteristic values ​​of the set gas smart valve.

[0012] Furthermore, the gas valve perimeter image data specifically refers to the row element value and two-dimensional coordinates of each pixel in the gas valve perimeter image, and the perimeter state recognition model includes an input preprocessing layer, a contour enhancement layer, a feature extraction layer, and an output layer.

[0013] Further, the specific steps for analyzing the perimeter anomaly response feature set of the set gas smart valve are as follows: In the input preprocessing layer of the perimeter state recognition model, the gas valve perimeter image data of the set gas smart valve is received and preprocessed; in the contour enhancement layer of the perimeter state recognition model, based on the preprocessed gas valve perimeter image data of the set gas smart valve, several pixel point sets of the set gas smart valve are extracted; in the feature extraction layer of the perimeter state recognition model, based on each pixel point set of the set gas smart valve, the perimeter anomaly feature vector of the set gas smart valve is extracted; in the output layer of the perimeter state recognition model, based on the perimeter anomaly feature vector of the set gas smart valve, the perimeter anomaly response feature set of the set gas smart valve is output.

[0014] Furthermore, the specific formula for calculating the pipeline transportation risk characteristic value of the set gas intelligent valve is as follows: ;in, , , The following are, in order, the characteristic values ​​for pipeline transportation risk of the set gas intelligent valve, the characteristic values ​​for abnormal gas transportation response, and the characteristic values ​​for perimeter disturbance degradation. The coefficients stored in the database are, in order: anomaly response coefficient, perimeter degradation coefficient, adjustment coefficient, cooperative smoothing coefficient, and gain coefficient. .

[0015] Furthermore, the specific steps for intelligent control of the set gas smart valve based on the pipeline transportation risk characteristic value are as follows: compare the pipeline transportation risk characteristic value of the set gas smart valve with the preset pipeline transportation risk characteristic range; and perform intelligent control of the set gas smart valve based on the comparison result.

[0016] The present invention has the following beneficial effects: (1) The gas smart valve based on narrowband IoT communication uses a two-level mechanism of data acquisition and analysis unit and local anomaly judgment unit to first analyze the abnormal response characteristics of gas transmission in real time and compare them with the threshold. Only when the threshold is exceeded is the narrowband communication transmission unit triggered to collect the valve perimeter image. The perimeter image analysis unit extracts the perimeter disturbance deterioration characteristics and merges them with the abnormal response characteristics of gas transmission to form the pipeline transmission risk characteristics. The intelligent feedback control unit implements a graded strategy according to the risk range, thereby realizing the distinguishing response to instantaneous fluctuations and continuous defects, thus significantly improving the precision of control, reducing the risk of missed reports, and reducing the frequency of operation and maintenance intervention while ensuring stable gas transmission, so as to reduce operating costs.

[0017] (2) The gas smart valve based on narrowband Internet of Things communication, through the data acquisition and analysis unit, comprehensively utilizes the gas transportation status data to decompose and form gas flow stability characteristic value and medium impact response characteristic value, and derives gas transportation abnormal response characteristic value, thereby comprehensively reflecting the real state of the transportation link under complex operating conditions. The gas transportation abnormal response characteristic value obtained therefrom can be directly used as the input for subsequent perimeter disturbance deterioration characteristic value fusion, thereby making the analysis of pipeline transportation risk characteristic value more complete and accurate, and providing a high-confidence decision basis for the intelligent feedback control unit, thereby achieving stable and safe operation guarantee for gas transportation.

[0018] (3) The gas smart valve based on narrowband IoT communication performs layered processing on the gas valve perimeter image data based on a pre-trained perimeter state recognition model. The input preprocessing layer is used for noise reduction, and the contour enhancement layer enhances the details of the gas valve perimeter image to highlight abnormal areas. The feature extraction layer generates corresponding perimeter abnormal feature vectors for different pixel sets and combines them in the output layer to form a perimeter abnormal response feature set. This can improve the stability of feature extraction while ensuring processing accuracy, and make the generated perimeter disturbance degradation feature values ​​more realistically reflect the structural state. This provides high-quality input for the accurate analysis of pipeline transport risk feature values, thereby improving the effectiveness of overall intelligent control.

[0019] Of course, any product implementing this invention does not necessarily need to achieve all of the advantages described above at the same time. Attached Figure Description

[0020] Figure 1 This is a block diagram of a gas smart valve based on narrowband Internet of Things communication according to the present invention.

[0021] Figure 2 This is a flowchart illustrating the specific steps involved in analyzing and setting the pipeline transport risk characteristic values ​​of a gas smart valve based on narrowband Internet of Things communication, according to the present invention.

[0022] Figure 3 This is a schematic diagram of the abnormal timing data of the perimeter abnormal response feature set of the gas intelligent valve in the present invention based on narrowband Internet of Things communication.

[0023] Figure 4 This is a flowchart illustrating the specific steps involved in analyzing and setting the perimeter anomaly response feature set of a gas smart valve based on narrowband Internet of Things communication according to the present invention. Detailed Implementation

[0024] Please see Figure 1 This invention provides a technical solution: a gas smart valve based on narrowband Internet of Things communication, comprising the following steps: a data acquisition and analysis unit, used to acquire in real time the gas delivery status data of the set gas smart valve (located in the gas pipeline section), and analyze the gas delivery abnormal response characteristic value of the set gas smart valve; a local anomaly determination unit, used to determine whether the gas delivery abnormal response characteristic value of the set gas smart valve is higher than a preset gas delivery abnormal response characteristic threshold. The narrowband communication transmission unit is used to acquire (when the gas delivery abnormal response characteristic value of the set gas intelligent valve is higher than the preset gas delivery abnormal response characteristic threshold) the gas valve perimeter image data of the set gas intelligent valve (at the preset image acquisition position, through a fixed or movable image acquisition device, the gas valve body and its adjacent pipe interfaces, flanges, welds, pipe surfaces, etc. within the visible range of the view are photographed to obtain a gas valve perimeter image data containing the visible areas of the above parts), and combine it with gas delivery status data (it should be explained that the timestamp of the acquisition is recorded when the gas valve perimeter image data and the gas delivery status data are acquired) and the corresponding gas delivery abnormal response characteristic value to generate a transmission data packet (specifically: the gas valve perimeter image data, gas delivery status data and gas delivery abnormal response characteristic value are combined according to a preset data encapsulation format to generate a transmission data packet, the encapsulation format contains seven fields arranged in order: data type identifier field, encoding rule such as: 0x01=image data segment, 0x02=status data segment, 0x03=characteristic value data segment; timestamp field, storing UNIX millisecond-level timestamp; unique identifier field (12 bytes) The data includes: a MAC address for the gas smart valve; an image data segment storing a lossless binary stream compressed using JPEG-LS with a compression ratio ≤10:1 and YUV420 sampling; a status data segment storing JSON structured text, where key-value pairs are defined as {"differential pressure": float, "temperature": float, "turbulence intensity": float, "moisture content": float, "vibration amplitude": float, "flow velocity pulsation": float}, with values ​​retained to 3 decimal places; and a gas delivery anomaly response characteristic value data segment storing a single 32-bit floating-point number. The data packet contains a gas delivery anomaly response characteristic value and an integrity verification field that stores a CRC-32 checksum. This field is generated by concatenating the binary values ​​of the image data segment, the gas delivery status data segment, and the gas delivery anomaly response characteristic value data segment (the data packet is generated by concatenating the fields in order). The data is then transmitted via a narrowband IoT network. (Note that if the gas delivery anomaly response characteristic value of the smart gas valve is set to be no higher than the preset gas delivery anomaly response characteristic threshold, the data packet is directly generated and transmitted via the narrowband IoT network. Then, the data packet at the current moment is saved.) The perimeter image analysis unit receives (transmission data packets delivered by the narrowband communication transmission unit and whose integrity has been verified) the transmission data packets according to a preset encapsulation format, and parses the transmission data packets to obtain gas valve perimeter image data, gas delivery status data, gas delivery abnormal response feature values, and corresponding timestamps and unique identifiers. Among them, the image data segment is decoded using JPEG-LS and converted into a standard image format for subsequent processing, the status data segment is parsed according to predefined JSON key-value pairs and maintains three decimal places of precision, and the abnormal response feature values ​​are read in 32-bit floating-point format; then, the image data and status data are time-aligned according to the timestamp) the gas valve perimeter image data and gas delivery abnormal response feature values ​​(and gas delivery status data) of the set gas smart valve, and analyzes the pipeline delivery risk feature values ​​of the set gas smart valve based on the pipeline delivery risk feature values. The intelligent feedback control unit is used to intelligently control the set gas smart valve based on the pipeline delivery risk feature values.

[0025] Specifically, the gas delivery status data includes pipe pressure difference, pipe gas temperature, pipe turbulence vortex intensity, pipe water vapor content, pipe wall vibration amplitude, and flow velocity pulsation amplitude. The specific steps for analyzing the abnormal response characteristic values ​​of the set gas intelligent valve are as follows: Based on the gas delivery status data of the set gas intelligent valve, analyze the delivery characteristic set of the set gas intelligent valve, including gas flow stability characteristic values ​​and medium impact response characteristic values; based on the delivery characteristic set of the set gas intelligent valve, analyze the abnormal response characteristic values ​​of the set gas intelligent valve, specifically by: weighting the gas flow stability characteristic values ​​and medium impact response characteristic values ​​of the set gas intelligent valve (and during this weighting process...). The weighting coefficients corresponding to the stable characteristic values ​​of gas flow and the impact response characteristic values ​​of the medium can be obtained through the following steps: obtain the stable characteristic values ​​of gas flow and the impact response characteristic values ​​of the medium at several historical time points, extract the mean value of the stable characteristic value of gas flow and the mean value of the impact response characteristic value of the medium respectively, sum them up to obtain the sum value of the abnormal response of gas transportation, and then compare the mean value of the stable characteristic value of gas flow and the mean value of the impact response characteristic value of the medium with the sum value of the abnormal response of gas transportation respectively. The results are used as the weighting coefficients corresponding to the stable characteristic value of gas flow and the impact response characteristic value of the medium respectively. In the weighting process, the stable characteristic value of gas flow is inverted, such as 1 / (1+stable characteristic value of gas flow).

[0026] The pipe pressure difference is the difference between the gas pressure before and after the valve. Its magnitude can reflect the resistance state and flow change trend of the gas when passing through the valve body or pipe section. It can be obtained by installing high-precision pressure sensors before and after the valve, collecting the instantaneous pressure values ​​at the two points, and performing differential processing to use the result as the pipe pressure difference value.

[0027] The gas temperature inside the pipe is the temperature of the gas inside the pipe, which can be obtained by deploying a high-temperature thermocouple inside the pipe.

[0028] The turbulent vortex intensity value inside the pipe is the instantaneous intensity of the vortex structure generated when the gas flows inside the pipe. This value can reflect the stability of the flow field inside the pipe and the degree of local turbulence. It can be obtained by deploying a Doppler ultrasonic velocity sensor array at the detection location in the pipe, acquiring the instantaneous flow velocity at multiple measuring points, and performing difference processing (taking the absolute value) on the instantaneous flow velocity of adjacent measuring points. Based on the result, the root mean square processing is performed, and the result is used as the turbulent vortex intensity value inside the pipe.

[0029] The water vapor content value in the pipe is the proportion of water vapor contained in the gas in the pipe. It can reflect the humidity status of the gas and the risk of condensation. It can be detected in real time by installing a high-sensitivity infrared absorption humidity sensor at the detection location of the pipe. The absorption of water vapor molecules in the gas by infrared light of a specific wavelength (such as 1.38μm) can be converted into a water vapor content value (based on the internal calibration curve of the sensor).

[0030] The vibration amplitude of the pipe wall is the instantaneous vibration displacement amplitude of the pipe wall under the action of gas flow. It can reflect the dynamic state of the pipe wall under fluid pulsation and mechanical disturbance. It can be obtained by deploying a high-frequency response acceleration sensor at the detection position on the outer wall of the pipe.

[0031] The velocity pulsation amplitude is the instantaneous change in gas velocity, which reflects the pulsation characteristics and pressure fluctuations during gas transportation. It can be obtained by deploying instantaneous response ultrasonic flow meters at pipeline detection locations to collect the instantaneous gas velocity value in real time and obtain the gas instantaneous velocity reference value (i.e., obtaining historical gas instantaneous velocity values ​​at several time points, averaging them, and using the result as the gas instantaneous velocity reference value). The gas instantaneous velocity value is then compared with the gas instantaneous velocity reference value, i.e., the absolute value of the difference between the gas instantaneous velocity value and the gas instantaneous velocity reference value / the gas instantaneous velocity reference value, and the result is used as the velocity pulsation amplitude.

[0032] The specific steps for analyzing and setting the delivery characteristic set of the smart gas valve are as follows: Based on the pipe pressure difference value, pipe turbulence vortex intensity value, and flow velocity pulsation amplitude of the smart gas valve, analyze the stable characteristic value of gas flow of the smart gas valve. Specifically, the pipe pressure difference value, pipe turbulence vortex intensity value, and flow velocity pulsation amplitude of the smart gas valve are standardized, and weighted processing is performed based on the standardization results (and in this weighting process, the weight coefficients corresponding to the standardized pipe pressure difference value, pipe turbulence vortex intensity value, and flow velocity pulsation amplitude can be obtained through the following...). Steps for obtaining standardized pipe pressure difference, turbulent vortex intensity, and velocity fluctuation amplitude values ​​at several historical time points are obtained. The mean values ​​of these three values ​​are extracted and summed to obtain a flow stability sum. The mean values ​​of these values ​​are then compared with the flow stability sum, and the results are used as weighting coefficients for the standardized pipe pressure difference, turbulent vortex intensity, and velocity fluctuation amplitude, respectively, to obtain the set gas pressure difference. The gas flow stability characteristic value of the smart valve (used to characterize the stability of the gas flow state in the gas pipeline section where the smart valve is located), and in the weighted processing, the standardized pipe pressure difference value, pipe turbulence vortex intensity value, and flow velocity pulsation amplitude are all reversed, such as 1 / (1+standardized pipe pressure difference value); based on the set gas temperature value, pipe water vapor content value, and pipe wall vibration amplitude of the smart valve, the medium impact response characteristic value of the set smart valve is analyzed, specifically: the set gas temperature value, pipe water vapor content value, and pipe wall vibration amplitude of the smart valve are... The pipe wall vibration amplitude is standardized, and a weighted processing is performed based on the standardization results (and in this weighted processing, the weighting coefficients corresponding to the standardized pipe gas temperature, pipe water vapor content, and pipe wall vibration amplitude are obtained in the same logical way as the standardized pipe pressure difference, pipe turbulent vortex intensity, and flow velocity pulsation amplitude). This is to obtain the medium impact response characteristic value of the set gas smart valve (used to characterize the gas medium thermal and moisture changes of the gas in the gas pipe section where the gas smart valve is located and the degree of structural dynamic response caused by them).

[0033] In this implementation scheme, by grouping and analyzing the gas transportation status data, and sequentially analyzing the gas flow stability characteristic value and the medium impact response characteristic value, a comprehensive quantification of the gas flow state and stability within the pipeline section, as well as the degree of thermal and moisture changes and structural dynamic response, is achieved. Secondly, standardization processing eliminates the differences in the dimensions of different physical quantities, ensuring the comparability of each parameter during analysis. Furthermore, the inversion of stability-related parameters during the weighting process can intuitively reflect the negative effect on flow stability when the value increases, ensuring that the characteristic values ​​are consistent with the actual physical meaning. Finally, the abnormal response characteristic value of gas transportation obtained in this way integrates the influence of both flow stability and medium impact, providing a high-precision input that can be directly used in the calculation when subsequently fused with the perimeter disturbance degradation characteristic value, thereby improving the accuracy and reliability of the pipeline transportation risk characteristic value analysis.

[0034] Specifically, the transmission steps based on the narrowband IoT network are as follows: A DTLS encrypted connection is established based on the NB-IoT modem (and the base station) (after handshake and key negotiation, the data transmission phase begins); the transmitted data packets are fragmented according to the CoAP protocol to obtain several fragmented data packets (with a single fragment length ≤ 1280 bytes). Specifically, according to the preset maximum fragment length of 1280 bytes, the transmitted data packets are sequentially truncated in byte order, and a CoAP header is appended to each fragment. The header includes a message type field (set to acknowledgment), a message code field, a message identifier field, and a fragment identifier field; the fragment identifier field records the fragment sequence number (starting from 1), the total number of fragments, and the last fragment flag (1 for the last fragment, 0 for other fragments). After completion, multiple fragmented data packets conforming to the CoAP protocol format (i.e., a CoAP fragmented message sequence) are obtained. Each fragmented data packet is transmitted piecemeal based on the UDP / IP protocol stack. Specifically, the fragmented data packets are transmitted sequentially by calling the UDP / IP protocol stack in ascending order of their fragmentation sequence number. A UDP datagram is constructed for each CoAP fragment: the source port field is set to 5683, the destination port field to 5684, the length field is the sum of the 8-byte UDP header length and the fragment payload length, and the checksum field is set to zero to disable UDP checksum (effective in IPv4 environment). Then, the UDP header and CoAP fragmented message are sequentially concatenated to form a UDP datagram, which is encrypted under an established DTLS session and transmitted via NB-IoT wireless bearer. When the UDP datagram length exceeds the current path MTU, IPv4 fragmentation processing is performed: the maximum payload length of a single IP fragment is determined according to the 8-byte alignment principle, and the original UDP datagram is sequentially divided into several consecutive payload blocks from the beginning. An IP header is added to each payload block, the identifier field replicates the message identifier field value of this message, and the fragment's identifier is calculated and filled in according to the above rules. The byte offset and corresponding fragment offset field (i.e., the result of dividing the byte offset by 8) are used, and the MF flag is set to 1 for non-last fragments and 0 for the last fragment. Each IP fragment is sent sequentially in the order of generation. The cloud receiving end (the server where the perimeter image analysis unit is located) aggregates the IP fragments according to the IP identifier field and reassembles the UDP datagram in ascending order according to the fragment offset field. After DTLS decryption, the original CoAP fragmented message is obtained, and then written to the corresponding position in the reassembly buffer according to the message identifier field and the fragment sequence number. When the fragment sequence numbers corresponding to the same message identifier from 1 to the total number of fragments are all present and the last fragment flag is valid, the data packet is reassembled in ascending order of fragment sequence number, and CRC-32 check is performed on its integrity check field. After successfully receiving each CoAP fragmented message, the receiving end returns an acknowledgment message (ACK), which carries the message identifier field, the current fragment sequence number, and the total number of fragments. After the entire packet is reassembled and the CRC-32 check passes, the receiving end returns a single overall acknowledgment message (ACK) containing the message identifier field, the last fragment sequence number, the total number of fragments, and the check pass flag.

[0035] In this implementation scheme, a DTLS encrypted connection is established between the NB-IoT modem and the base station to achieve secure encryption of the data transmission link of the gas smart valve. This ensures that gas delivery status data and perimeter image data are not stolen or tampered with during transmission over the public network. Secondly, a CoAP protocol fragmentation mechanism is adopted to fragment the transmission data packet into fragments with a limit of no more than 1280 bytes and attach protocol header information, clearly recording the fragment sequence number, total number of fragments, and last fragment flag. This ensures that the receiving end can accurately identify and recover the original data packet. Finally, in the UDP / IP protocol stack transmission stage, the ascending order of fragment sequence number and IPv4 fragmentation processing ensure that large-size data can still be transmitted completely under path MTU constraints. Efficient reassembly is achieved by combining the IP header identifier and fragment offset field. After reassembly, the receiving end performs CRC-32 integrity verification and provides reception confirmation for each fragment and the entire packet based on the ACK mechanism, thereby achieving reliable transmission of data packets and effectively supporting the real-time data interaction needs of the gas smart valve in a narrowband IoT environment.

[0036] Specifically, such as Figure 2 As shown, the specific steps for analyzing the pipeline transportation risk characteristics of the set gas smart valve are as follows: Based on the pre-trained perimeter state recognition model and combined with the gas valve perimeter image data of the set gas smart valve, analyze the perimeter disturbance degradation characteristics of the set gas smart valve; read the gas transportation abnormal response characteristics of the set gas smart valve and, combined with the perimeter disturbance degradation characteristics, analyze the pipeline transportation risk characteristics of the set gas smart valve.

[0037] The specific steps for analyzing the perimeter disturbance degradation characteristic values ​​of the set gas smart valve are as follows: Input the gas valve perimeter image data of the set gas smart valve into the pre-trained perimeter state recognition model, analyze the perimeter abnormal response feature set of the set gas smart valve, including surface abnormal seepage feature values, perimeter seal fracture disturbance feature values, and structural deformation trace feature values; Based on the perimeter abnormal response feature set of the set gas smart valve, analyze the perimeter disturbance degradation characteristic values ​​of the set gas smart valve.

[0038] The specific formula for calculating the perimeter disturbance degradation characteristic value of the set gas intelligent valve is as follows: ;in, To set the perimeter disturbance degradation characteristic value of the gas intelligent valve, To set the surface abnormal seepage characteristic value of the gas smart valve, The seepage coefficient is stored in the database. To set the characteristic value of the perimeter seal fracture disturbance of the gas smart valve, The fracture coefficient is stored in the database. To set the structural deformation trace characteristic values ​​of the gas intelligent valve, These are the deformation coefficients stored in the database. Furthermore, in this embodiment, the seepage coefficient stored in the database Fracture coefficient Deformation coefficient The values ​​were 0.438, 0.315, and 0.247, respectively.

[0039] It needs to be explained that the seepage coefficient stored in the database Fracture coefficient Deformation coefficient The specific acquisition steps are as follows: Obtain the surface abnormal seepage characteristic values, perimeter seal fracture disturbance characteristic values, and structural deformation trace characteristic values ​​of the set gas smart valve at several historical time points. Extract the mean values ​​of the surface abnormal seepage characteristics, perimeter seal fracture disturbance characteristics, and structural deformation trace characteristics respectively, and perform summation analysis to obtain the disturbance degradation sum value. Ratio the mean values ​​of the surface abnormal seepage characteristics, perimeter seal fracture disturbance characteristics, and structural deformation trace characteristics to the disturbance degradation sum value, and use the corresponding results as the seepage coefficient. Fracture coefficient Deformation coefficient .

[0040] The specific formula for calculating the pipeline transportation risk characteristic value of the set gas intelligent valve is as follows: ;in, To set the pipeline transportation risk characteristic value for the gas intelligent valve, To set the abnormal gas delivery response characteristic value of the gas intelligent valve, These are the abnormal response coefficients stored in the database. To set the perimeter disturbance degradation characteristic value of the gas intelligent valve, The perimeter degradation coefficient is stored in the database. These are adjustment coefficients stored in the database. These are the collaborative smoothing coefficients stored in the database. The gain coefficients are stored in the database. Furthermore, in this implementation example, the abnormal response coefficients stored in the database Perimeter degradation coefficient Adjustment coefficient Cooperative smoothing coefficient Gain coefficient The values ​​were 0.543, 0.457, 2.000, 0.316, and 2.000, respectively.

[0041] It needs to be explained that the abnormal response coefficients stored in the database Perimeter degradation coefficient The acquisition steps are as follows: Obtain the abnormal gas delivery response characteristic values ​​and perimeter disturbance degradation characteristic values ​​of the set gas smart valve at several historical time points. Extract the mean values ​​of the abnormal gas delivery response characteristics and perimeter disturbance degradation characteristics respectively, and perform summation analysis to obtain the delivery risk sum value. Ratio the mean values ​​of the abnormal gas delivery response characteristics and perimeter disturbance degradation characteristics to the delivery risk sum value, and use the corresponding results as the abnormal response coefficient. Perimeter degradation coefficient .

[0042] Adjustment coefficient Gain coefficient All are preset constants, with a value set to 2.000 during the initialization phase. These constants are used to maintain a balance between response stability and risk regulation during the gas delivery process, and also serve as a cooperative smoothing coefficient. The acquisition steps are as follows: Obtain the abnormal gas delivery response characteristic values ​​and perimeter disturbance degradation characteristic values ​​of the set gas smart valve at several historical time points, extract the correlation value between the two based on the Pearson correlation coefficient, and use it as the cooperative smoothing coefficient. .

[0043] The specific implementation example for calculating the pipeline transportation risk characteristic value of the set gas smart valve is as follows. The existing data includes the surface abnormal seepage characteristic value, perimeter seal fracture disturbance characteristic value, and structural deformation trace characteristic value for any 5 abnormal time points of the set gas smart valve (i.e., the time points corresponding to the gas transportation abnormal response characteristic value of the set gas smart valve at any 5 time points being higher than the preset gas transportation abnormal response characteristic threshold, which are marked as abnormal time points). The acquisition steps can be achieved by obtaining the gas valve perimeter image data at the corresponding abnormal time points and inputting it into the perimeter state recognition model. This model processes the data sequentially through an input preprocessing layer, a contour enhancement layer, a feature extraction layer, and an output layer, ultimately obtaining the surface abnormal seepage characteristic value, perimeter seal fracture disturbance characteristic value, and structural deformation trace characteristic value for the corresponding abnormal time points in the output layer. See Table 1 for details. Figure 3 As shown: Table 1. Example of abnormal time-series data for setting the perimeter abnormal response feature set of a gas intelligent valve.

[0044] Seepage coefficients stored in the database The value is: 0.438; Fracture coefficients stored in the database The value is: 0.315; Deformation coefficients stored in the database The value is: 0.247; Substituting the data from Table 1 and the aforementioned coefficients into the specific formula for calculating the perimeter disturbance degradation characteristic value of the set gas intelligent valve, we obtain: The perimeter disturbance degradation characteristic value of the first abnormal time point of the gas intelligent valve is set to 0.438×0.259+0.315×0.346+0.247×0.28≈0.293; The perimeter disturbance degradation characteristic value of the second abnormal time point of the gas intelligent valve is set to 0.438×0.342+0.315×0.416+0.247×0.364≈0.371; The perimeter disturbance degradation characteristic value of the third abnormal time point of the gas intelligent valve is set to 0.438×0.487+0.315×0.512+0.247×0.443≈0.484; The perimeter disturbance degradation characteristic value of the fourth abnormal time point of the gas intelligent valve is set to 0.438×0.542+0.315×0.568+0.247×0.484≈0.536; The perimeter disturbance degradation characteristic value of the fifth abnormal time point of the gas intelligent valve is set to 0.438×0.583+0.315×0.612+0.247×0.447≈0.559.

[0045] In this implementation plan, by fusing the abnormal response characteristic values ​​of gas transportation with the perimeter disturbance deterioration characteristic values ​​extracted based on the perimeter state identification model, a comprehensive quantification of pipeline transportation risk is achieved. Secondly, the perimeter disturbance deterioration characteristic values ​​consist of surface abnormal seepage characteristic values, perimeter seal fracture disturbance characteristic values, and structural deformation trace characteristic values, and are weighted using corresponding coefficients stored in the database. This ensures that the impact of different types of structural defects is reasonably reflected in the total characteristic value. Finally, the analysis of multiple coefficients for pipeline transportation risk characteristic values ​​uses a formulaic fusion method to uniformly represent the risk contribution of internal transportation status and external structural status. This avoids bias caused by a single indicator dominating the judgment, and ensures that the final risk characteristic value reflects both the degree of decline in transportation stability and the potential threat of perimeter structural deterioration to transportation safety, thus providing accurate quantitative basis for intelligent feedback control.

[0046] Specifically, please refer to Figure 4 The gas valve perimeter image data specifically consists of the row element value and two-dimensional coordinates of each pixel in the gas valve perimeter image, and the perimeter state recognition model includes an input preprocessing layer, a contour enhancement layer, a feature extraction layer, and an output layer.

[0047] The specific steps for analyzing the perimeter anomaly response feature set of the set gas smart valve are as follows: In the input preprocessing layer of the perimeter state recognition model, the gas valve perimeter image data of the set gas smart valve is received and preprocessed, such as noise suppression processing. Through the joint operation of Gaussian filtering and adaptive median filtering, high-frequency noise points and isolated bright spots caused by air particle reflection are eliminated respectively, while retaining detailed pixel information such as sealing line boundary, seepage traces and deformation contours; In the contour enhancement layer of the perimeter state recognition model, based on the preprocessed gas valve perimeter image data of the set gas smart valve, several pixel point sets of the set gas smart valve are extracted. Specifically, this involves calling a pre-set existing contour segmentation network based on a deep convolutional neural network (such as the improved DeepLab). The v3+ or U-Net structure performs pixel-by-pixel classification and annotation on pixels in the image corresponding to the gas valve perimeter sealing line, pipe interface and its adjacent areas. During the training phase, this segmentation network has been supervised learning through labeled samples containing multiple types of seepage traces, crack boundaries, seal line breaks and structural deformation traces. It can output the category label and confidence value of each pixel during the inference phase. Based on the output results, the pixel sets belonging to the seepage trace, seal line break and deformation trace categories are selected and stored as surface abnormal seepage pixel set (including the pixel value and two-dimensional coordinates of each seepage trace pixel), perimeter sealing line defect pixel set (including the pixel value and two-dimensional coordinates of each seal line break pixel), and structural deformation trace pixel set (including the pixel value and two-dimensional coordinates of each deformation trace pixel), to obtain several pixel sets. In the feature extraction layer of the perimeter state recognition model, the perimeter anomaly feature vector of the set gas smart valve is extracted based on each pixel set of the set gas smart valve. In the output layer of the perimeter state recognition model, the perimeter anomaly response feature set of the set gas smart valve is output based on the perimeter anomaly feature vector of the set gas smart valve. Specifically, the surface anomaly seepage feature, perimeter seal fracture disturbance feature, and structural deformation trace feature in the perimeter anomaly feature vector are activated by the Sigmoid function, and the results are mapped between 0 and 1 to obtain the surface anomaly seepage feature value, perimeter seal fracture disturbance feature value, and structural deformation trace feature value of the set gas smart valve.

[0048] The specific steps for extracting the perimeter anomaly feature vector of the set gas smart valve are as follows: Clustering of connected components (using 8-neighborhood) was performed on the surface of the gas smart valve to identify abnormal seepage pixels, resulting in several seepage connected components. A covariance matrix was constructed based on the two-dimensional coordinates of each seepage trace pixel within each seepage connected component, and principal component analysis (PCA) was performed to extract the maximum and minimum eigenvalues. The ratio of these eigenvalues ​​was calculated as (maximum eigenvalue - minimum eigenvalue) / (maximum eigenvalue + minimum eigenvalue) to determine the fineness of each seepage connected component. The eigendirection vector corresponding to the maximum eigenvalue was used as the principal direction angle of the corresponding seepage connected component. Each principal direction angle was divided into several direction intervals, and the number of seepage connected components contained in each direction interval was counted. The most frequent directional interval is identified, and the mean of the principal directional angle of each seepage connected domain within this interval is extracted as the reference directional angle. The principal directional angle of each seepage connected domain is then compared with the reference directional angle by the difference, i.e., 1-[wrap(principal directional angle - reference directional angle) / π]. wrap() represents a function that reduces the angle difference to [-π, π] to obtain the directional difference characteristics of each seepage connected domain. The difference is then standardized with the fineness of the filaments. Based on the standardized results, a weighted average is applied to extract abnormal seepage features on the surface (used to characterize the degree of abnormality of surface seepage traces, and to quantify the degree of abnormality in the formation and continuation of seepage traces; the larger the value, the more obvious and stable the seepage traces, and the higher the degree of abnormality). The system reads the two-dimensional coordinates of each broken pixel in the perimeter sealing line defect pixel set of the set gas smart valve, and performs connected component clustering (using the 8-neighborhood judgment rule) to divide the spatially connected defect pixels into several independent defect connected components (each defect connected component corresponds to a local break or damaged segment of the sealing line). For each broken pixel in each defect connected component, a morphological opening operation is performed (the structuring element radius is 2 to 3 pixels) to perform spatial noise reduction processing, removing isolated bright spots and single-pixel pseudo-connections around the sealing line. Then, the Zhang-Suen thinning algorithm is used to obtain the skeleton line of each defect connected component (the skeleton line is composed of all the broken pixels in the sealed line retained after thinning in the defect connected component, and each is marked as a skeleton pixel). Each skeleton pixel of the skeleton line of each defect connected component is topologically sorted according to the connection order, that is, its coordinate order is recorded point by point along the connected skeleton, and the closure is guaranteed under the condition that the difference between the coordinates of the starting point and the ending point is within 1 pixel, so as to obtain the sorted skeleton polyline of each defect connected component (that is, including several sorted skeleton pixels). The Euclidean distances between adjacent skeleton pixels are analyzed and accumulated sequentially (i.e., based on the two-dimensional coordinate analysis of the sorted skeleton pixels) to obtain the geodesic length of the skeleton connected region of the defect. The two-dimensional coordinates of each pair of adjacent skeleton pixels in the sorted skeleton polyline are then read sequentially and recorded as the current skeleton pixel and the next skeleton pixel, respectively. Based on the coordinate difference between these two points, the orientation angle value of the skeleton line in that segment is analyzed. Subsequently, the orientation angles of adjacent segments in the sorted skeleton polyline are differentially calculated to obtain the angle change value at that point. The angle change value is limited to the range of -180 degrees to +180 degrees to avoid the influence of orientation jumps caused by angle accumulation. This process is then used to obtain the orientation change value in the sorted skeleton polyline. After considering all adjacent corner changes, the absolute values ​​of these corner changes are summed and divided by the total number of segments of the sorted skeleton polyline minus two. The result is then proportionally converted to a normalized range of zero to one to obtain the irregularity value of the skeleton line. When the number of skeleton pixels of the sorted skeleton polyline is less than three, the irregularity value is set to zero and standardized with the geodesic length of the skeleton. The standardized result is then weighted and averaged to extract the perimeter seal fracture disturbance features (used to characterize the degree of abnormality of the perimeter seal fracture on the overall stability, and to quantify the degree of damage to the seal continuity; the larger the value, the more significant the fracture impact and the higher the degree of abnormality). The system reads the two-dimensional coordinates of each deformation trace pixel in the set of structural deformation trace pixels of the gas intelligent valve, and performs region division processing (such as 8-neighborhood) to obtain several deformation regions. The pixel value of each deformation trace pixel in each deformation region is processed into grayscale to obtain the corresponding grayscale pixel value. The two-dimensional coordinates of each deformation trace pixel in each deformation region are then averaged to extract the two-dimensional coordinates of the centroid of the corresponding deformation region. Based on the Euclidean distance formula, the Euclidean distance from each deformation trace pixel in each deformation region to the centroid's two-dimensional coordinates is analyzed and multiplied by its corresponding grayscale pixel value. The result is then averaged to obtain the radial energy distribution value of the deformation region. Then, the two-dimensional coordinates of any two adjacent deformation trace pixels within the deformation region are sequentially selected, and the direction angle of the line connecting the two deformation trace pixels relative to the horizontal direction is analyzed based on the arctangent function. This process is then used to calculate the radial energy distribution value of all adjacent deformation trace pixels within the deformation region. The standard deviation of the orientation angle set of the deformation trace pixel pairs is calculated to obtain the orientation angle dispersion of the deformation region. For the grayscale pixel value of each deformation trace pixel in the deformation region, the horizontal and vertical gradient components are calculated separately based on the Sobel operator. The gradient magnitude of each deformation trace pixel is calculated based on the square root formula and then processed to obtain the standard deviation of the gradient magnitude of the deformation region. This is then standardized with the radial energy distribution value and the orientation angle dispersion. The standardized results are weighted and mean-valued to extract structural deformation trace features (used to characterize the degree of structural deformation anomaly, and to quantify the comprehensive anomaly degree of deformation in spatial distribution, orientation stability and boundary uniformity; the larger the value, the more significant the deformation and the higher the degree of anomaly). The surface abnormal seepage features, the perimeter sealing fracture disturbance features and the structural deformation trace features are concatenated into a perimeter anomaly feature vector.

[0049] The pre-training steps for the perimeter state recognition model are as follows: A dataset of annotated images of the gas valve perimeter was obtained, including several sets of gas valve perimeter image samples and their corresponding perimeter anomaly category labels. The categories include abnormal surface seepage, perimeter seal breakage disturbance, structural deformation traces, and normal areas. The row element value, two-dimensional coordinates, and anomaly category label of each pixel were annotated in the samples. At the same time, the boundary contour, pixel set information, and morphological feature description of each anomaly region were recorded in the annotation file. The annotated dataset was divided into training set and validation set according to the proportion to ensure that the distribution of samples of each category is balanced.

[0050] The perimeter state recognition model is structurally initialized, which involves initializing the weight parameters and bias terms of each convolutional and fully connected layer in the input preprocessing layer, contour enhancement layer, feature extraction layer, and output layer. Xavier initialization or He initialization strategies are used to ensure a reasonable distribution of initial weight values, and the bias terms are initialized to zero. The activation function is ReLU or LeakyReLU to enhance the model's nonlinear feature extraction capability, and the output layer of the loss function uses Sigmoid activation to ensure that feature values ​​are mapped to the 0-1 interval.

[0051] Pre-training is performed based on the training set, with a set number of training loops. The following steps are executed in each training loop: Forward propagation: The gas valve perimeter image samples are input into the model, sequentially passing through the noise suppression module (Gaussian filtering and adaptive median filtering) of the input preprocessing layer, and the convolution-pooling structure of the contour enhancement layer (calling a pre-defined improved DeepLab v3+ or U-Net segmentation network). In the feature extraction layer, pixel feature vectors of seepage traces, seal line breakage, and deformation traces are extracted. Surface abnormal seepage feature values, perimeter seal breakage disturbance feature values, and structural deformation trace feature values ​​are generated in the output layer. Loss function calculation: A multi-objective joint loss function is used, including pixel semantic segmentation loss (cross-entropy loss) and abnormal feature value regression loss (mean squared error), comprehensively measuring the matching degree between the model output and the labeled data. Backpropagation and parameter update: Backpropagation is performed based on the loss function results, calculating the gradient values ​​of the weights and biases of each layer, and using the Adam optimizer to update the network parameters to minimize the training error.

[0052] After each training cycle, a validation set is used for performance evaluation to verify forward propagation: the validation set images are input into the model, and the predicted perimeter anomaly feature values ​​and category distribution map are output. Performance evaluation metrics are calculated: based on the validation set annotations, pixel classification accuracy, mIoU (Mean Intersection over Union), recall, and mean squared error of feature values ​​are calculated. Hyperparameter adjustment and convergence judgment: if the validation set loss does not decrease or the accuracy does not improve within several consecutive cycles, the learning rate, network depth, regularization coefficient, and other hyperparameters are adjusted. When the validation set metrics reach the preset standard or the loss function does not decrease significantly for several consecutive cycles, the training process is stopped to prevent overfitting.

[0053] Ultimately, a trained perimeter state recognition model was obtained, which met the design requirements in terms of perimeter anomaly recognition accuracy, feature value extraction stability, and anomaly pattern adaptability.

[0054] In this implementation scheme, the joint operation of Gaussian filtering and adaptive median filtering effectively suppresses high-frequency noise and isolated bright spots, preserves key details, and provides high-quality input for subsequent contour segmentation. Secondly, with the help of improved DeepLab v3+ or U-Net deep segmentation structures, pixel-by-pixel accurate segmentation of multiple anomalies is achieved to ensure the spatial positioning accuracy of feature extraction. Furthermore, the feature extraction layer performs detailed analysis on different types of pixel sets, such as the filamentity and directional differences of seepage connectivity, the irregularity and geodesic length of sealing skeleton lines, and the radial energy and gradient distribution of deformation regions. This ensures that the output features not only reflect the existence of anomalies but also quantify the degree of anomalies. Finally, the feature values ​​are standardized to the 0-1 range through Sigmoid mapping, making it easy to directly call them in subsequent risk assessment or alarm strategies, achieving seamless integration of data and decision-making, thereby improving the overall stability and accuracy of identification.

[0055] Specifically, the steps for intelligent control of the set gas smart valve based on the pipeline transportation risk characteristic value are as follows: compare the pipeline transportation risk characteristic value of the set gas smart valve with the preset pipeline transportation risk characteristic range; and perform intelligent control of the set gas smart valve based on the comparison result, specifically: if the pipeline transportation risk characteristic value of the set gas smart valve is lower than the lower limit of the preset pipeline transportation risk characteristic range, it is determined that the current risk is at an acceptable level, no shut-off control action is executed, the existing valve opening is maintained, and only status recording and routine reporting are performed. If the pipeline transportation risk characteristic value of the smart gas valve is set to be within the preset pipeline transportation risk characteristic range, it is judged as a medium-level risk, and steady-state protection control is executed (i.e., a steady-state protection control command is issued to the smart gas valve based on a narrowband IoT network). This means that an upper limit constraint is imposed on the valve opening (e.g., when the pipeline transportation risk characteristic value is close to the lower limit of the pipeline transportation risk characteristic range, the upper limit of the valve's allowed target opening is no higher than 80% of the preset rated opening; when the pipeline transportation risk characteristic value is located in the middle of the pipeline transportation risk characteristic range, the upper limit of the valve's allowed target opening is no higher than 70% of the preset rated opening; when the pipeline transportation risk characteristic value is close to the lower limit of the pipeline transportation risk characteristic range, the upper limit of the valve's allowed target opening is no higher than 70% of the preset rated opening). When the upper limit of the characteristic range is reached, the upper limit of the target opening of the valve shall not be higher than 60% of the preset rated opening. It should be noted that when approaching the lower limit of the pipeline transportation risk characteristic range, the value shall not be less than the lower limit of the range and not greater than the lower limit of the range plus 30% of the range width. When located in the middle of the pipeline transportation risk characteristic range, the value shall be greater than the lower limit of the range plus 30% of the range width and not greater than the upper limit of the range minus 30% of the range width. When approaching the upper limit of the pipeline transportation risk characteristic range, the value shall be greater than the upper limit of the range minus 30% of the range width and not greater than the upper limit of the range. At the same time, the warning flag shall be set (that is, the medium-level risk status shall be set to the effective status) and reported. If the pipeline transportation risk characteristic value of the set gas smart valve is higher than the upper limit of the preset pipeline transportation risk characteristic range, it is judged as a high-level risk and emergency protection control is executed (i.e., emergency protection control command is issued to the gas smart valve based on the narrowband Internet of Things network). The valve is immediately driven to close to the safe position (closed or the opening degree is reduced to below the preset safe opening degree threshold), and linkage commands such as pressure reduction / bypass cut-off are issued. High-level alarms are triggered simultaneously and reported through the communication network. When communication is interrupted, the local disconnection protection strategy is executed (i.e., the default is to enter the safe position).

[0056] This implementation plan establishes a hierarchical intelligent control mechanism based on quantified risk characteristic values. This enables the gas intelligent valve to execute precise control strategies at different risk levels. Specifically, by comparing the risk characteristic value of the pipeline segment with a preset range, the risk level is automatically determined, improving the sensitivity of risk identification. Secondly, for medium-level risks, the valve opening upper limit is set in different ranges, which not only dynamically adjusts the delivery capacity but also effectively reduces the potential hazards caused by pipeline pressure fluctuations. At the same time, a warning sign is set and reported, allowing maintenance personnel to monitor the risk status in real time. Finally, for high-level risks, emergency protection control is directly triggered, closing the valve to a safe position and implementing linked pressure reduction / cut-off measures. This ensures that the spread of the accident is quickly stopped in extreme situations, thereby significantly improving the safety of the gas pipeline network under different operating scenarios.

[0057] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0058] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A smart gas valve based on narrowband Internet of Things (IoT) communication, characterized in that, include: The data acquisition and analysis unit is used to acquire the gas delivery status data of the set gas smart valve in real time, and analyze the abnormal response characteristic values ​​of the set gas smart valve. The local anomaly determination unit is used to determine whether the gas delivery anomaly response characteristic value of the set gas smart valve is higher than the preset gas delivery anomaly response characteristic threshold. The narrowband communication transmission unit is used to acquire the gas valve perimeter image data of the set gas smart valve, combine the gas delivery status data and the corresponding gas delivery abnormal response characteristic value to generate a transmission data packet, and transmit it based on the narrowband Internet of Things network. The perimeter image analysis unit is used to receive the perimeter image data of the gas valve and the abnormal response feature value of gas transportation of the set gas smart valve, and analyze the pipeline transportation risk feature value of the set gas smart valve based on the pre-trained perimeter state recognition model. The intelligent feedback control unit is used to intelligently control the set gas intelligent valve based on the risk characteristics of pipeline transportation.

2. The gas smart valve based on narrowband IoT communication as described in claim 1, characterized in that, The gas delivery status data includes the pipe pressure difference value, the gas temperature value, the turbulent vortex intensity value, the water vapor content value, the pipe wall vibration amplitude, and the flow velocity pulsation amplitude. The specific steps for analyzing and setting the abnormal response characteristic values ​​of the gas delivery smart valve are as follows: Based on the gas delivery status data of the set gas intelligent valve, the delivery characteristic set of the set gas intelligent valve is analyzed, including gas flow stability characteristic value and medium impact response characteristic value. Based on the set delivery feature set of the set gas intelligent valve, the abnormal response feature value of the gas delivery of the set gas intelligent valve is analyzed.

3. The gas smart valve based on narrowband IoT communication as described in claim 2, characterized in that, The specific steps for analyzing and setting the delivery characteristic set of the gas intelligent valve are as follows: Based on the pipe pressure difference value, pipe turbulence vortex intensity value, and flow velocity pulsation amplitude of the set gas intelligent valve, the stable characteristic value of gas flow of the set gas intelligent valve is analyzed. Based on the set gas temperature, water vapor content, and pipe wall vibration amplitude of the intelligent gas valve, the media impact response characteristics of the intelligent gas valve are analyzed.

4. The gas smart valve based on narrowband Internet of Things communication as described in claim 1, characterized in that, The specific steps for transmission based on narrowband IoT networks are as follows: Establish a DTLS encrypted connection based on an NB-IoT modem; The transmitted data packets are fragmented based on the CoAP protocol to obtain several fragmented data packets; Each data packet is transmitted piece by piece based on the UDP / IP protocol stack.

5. The gas smart valve based on narrowband Internet of Things communication as described in claim 1, characterized in that, The specific steps for analyzing and setting the pipeline transportation risk characteristic values ​​for the gas intelligent valve are as follows: Based on a pre-trained perimeter state recognition model and combined with the perimeter image data of the set gas smart valve, the perimeter disturbance degradation characteristic value of the set gas smart valve is analyzed. Read the abnormal response characteristic value of the gas delivery of the set gas intelligent valve, and combine it with the perimeter disturbance deterioration characteristic value to analyze the pipeline delivery risk characteristic value of the set gas intelligent valve.

6. The gas smart valve based on narrowband Internet of Things communication as described in claim 5, characterized in that, The specific steps for analyzing and setting the perimeter disturbance degradation characteristic value of the gas intelligent valve are as follows: The perimeter image data of the set gas smart valve is input into the pre-trained perimeter state recognition model to analyze the perimeter abnormal response feature set of the set gas smart valve, including surface abnormal seepage feature value, perimeter seal fracture disturbance feature value, and structural deformation trace feature value. Based on the perimeter abnormal response feature set of the set gas intelligent valve, the perimeter disturbance degradation characteristic value of the set gas intelligent valve is analyzed.

7. The gas smart valve based on narrowband IoT communication as described in claim 6, characterized in that, The gas valve perimeter image data specifically refers to the row element value and two-dimensional coordinates of each pixel in the gas valve perimeter image, and the perimeter state recognition model includes an input preprocessing layer, a contour enhancement layer, a feature extraction layer, and an output layer.

8. The gas smart valve based on narrowband Internet of Things communication as described in claim 7, characterized in that, The specific steps for analyzing and setting the perimeter anomaly response feature set of the gas intelligent valve are as follows: In the input preprocessing layer of the perimeter state recognition model, the gas valve perimeter image data of the set gas smart valve is received and preprocessed. In the contour enhancement layer of the perimeter state recognition model, based on the preprocessed gas valve perimeter image data of the set gas smart valve, several pixel point sets of the set gas smart valve are extracted. In the feature extraction layer of the perimeter state recognition model, the perimeter anomaly feature vector of the set gas smart valve is extracted based on each pixel set of the set gas smart valve. In the output layer of the perimeter state recognition model, based on the perimeter anomaly feature vector of the set gas smart valve, the perimeter anomaly response feature set of the set gas smart valve is output.

9. The gas smart valve based on narrowband Internet of Things communication as described in claim 5, characterized in that, The specific formula for calculating the pipeline transportation risk characteristic value of the set gas intelligent valve is as follows: ； in, , , The following are, in order, the characteristic values ​​for pipeline transportation risk of the set gas intelligent valve, the characteristic values ​​for abnormal gas transportation response, and the characteristic values ​​for perimeter disturbance degradation. The coefficients stored in the database are, in order: anomaly response coefficient, perimeter degradation coefficient, adjustment coefficient, cooperative smoothing coefficient, and gain coefficient. .

10. The gas smart valve based on narrowband Internet of Things communication as described in claim 9, characterized in that, The specific steps for intelligent control of the gas smart valve based on the risk characteristics of pipeline transportation are as follows: The risk characteristic values ​​of the pipeline transportation of the set gas intelligent valve are compared with the preset risk characteristic range of the pipeline transportation. The gas intelligent valve is intelligently controlled based on the comparison and processing results.

Images