Identifying methane clusters and sources

Abstract

Systems, devices, and methods for detecting leaks near an emission source are disclosed, the methods comprising: (i) receiving a first information stream indicative of a leak condition from one or more moving sensors, (ii) determining that an initial leak condition exists based at least in part on the first information stream indicative of the leak condition, (iii) receiving a second information stream indicative of a non-leak condition, (iv) determining, using a statistical model, that the leak condition has ended based at least in part on the first information stream and the second information stream, (v) receiving a third information stream indicative of the leak condition, and (vi) determining that a new leak condition exists, the new leak condition being a different leak condition from the initial leak condition.

Description

Cross-reference to other applications

[0001] This application claims priority to U.S. Provisional Patent Application No. 63 / 415,223, entitled "LEAK LIFECYLE ANALYSIS," filed October 11, 2022, which is incorporated herein by reference for all purposes, and to U.S. Provisional Patent Application No. 63 / 442,910, entitled "METHANE SOURCE INDICATOR," filed February 2, 2023, which is incorporated herein by reference for all purposes, and to U.S. Provisional Patent Application No. 63 / 532,021, entitled "METHANE SOURCE INDICATOR," filed August 10, 2023, which is incorporated herein by reference for all purposes. [Background technology]

[0002] Monitoring environmental conditions measures the levels of various environmental components, enabling the detection of potentially harmful air pollution, radiation, greenhouse gases, or other hazardous substances in the environment. Depending on the application, environmental monitoring systems can be used in outdoor or indoor environments. Monitoring environmental conditions typically involves collecting environmental data. Environmental data includes the detection and measurement of pollutants or hazardous substances, such as nitrogen dioxide (NO), carbon monoxide (CO), nitric oxide (NO), ozone (O), sulfur dioxide (SO), carbon dioxide (CO), methane (CH), volatile organic compounds (VOCs), air toxics, temperature, sound emissions, and particulate matter. To assess the impact of such pollutants, it is desirable to associate environmental data detecting these pollutants at specific times with geographic locations (e.g., homes, businesses, towns, etc.). Such associations allow individuals and communities to assess the quality of their environments. Therefore, it is desirable for collected data to be representative of an area. Furthermore, it is desirable for the collected data to meet a desired error tolerance and be efficiently collected and processed. Therefore, mechanisms for improving the collection and processing of environmental data are desirable. [Brief explanation of the drawings]

[0003] Various embodiments of the present invention are disclosed in the following detailed description and the accompanying drawings.

[0004] [Figure 1] FIG. 1 illustrates one embodiment of a system for acquiring environmental data using a mobile sensor platform and associating the environmental data with map features.

[0005] [Figure 2] FIG. 1 illustrates an embodiment of a method for acquiring environmental data using a mobile sensor platform.

[0006] [Figure 3A] 1 illustrates a particular area and an embodiment of a route that may be traversed using a method for acquiring environmental data using a mobile sensor platform. [Figure 3B] 1 illustrates a particular area and an embodiment of a route that may be traversed using a method for acquiring environmental data using a mobile sensor platform. [Figure 3C] 1 illustrates a particular area and an embodiment of a route that may be traversed using a method for acquiring environmental data using a mobile sensor platform.

[0007] [Figure 4] FIG. 10 illustrates the relationship between methane and ethane signals in collected sensor data in accordance with various embodiments.

[0008] [Figure 5A] FIG. 1 illustrates an example of calibration of a set of sensors, according to various embodiments.

[0009] [Figure 5B] 4A-4C illustrate examples of moving baseline signals according to a first time window for methane and ethane signals in accordance with various embodiments. [Figure 5C]4A-4C illustrate examples of moving baseline signals according to a first time window for methane and ethane signals in accordance with various embodiments.

[0010] [Figure 5D] 6A-6C show examples of moving baseline signals according to a second time window for methane and ethane signals in accordance with various embodiments. [Figure 5E] 6A-6C show examples of moving baseline signals according to a second time window for methane and ethane signals in accordance with various embodiments.

[0011] [Figure 5F] FIG. 10 illustrates an example of an ethane signal relative to various baselines in accordance with various embodiments.

[0012] [Figure 6A] FIG. 1 illustrates an example of a map containing methane signals from different source types, according to various embodiments. [Figure 6B] FIG. 1 illustrates an example of a map containing methane signals from different source types, according to various embodiments.

[0013] [Figure 7] 1 illustrates a method for calibrating a sensor according to various embodiments.

[0014] [Figure 8] FIG. 1 illustrates a method for determining acquisition of different contaminant signals in collected data collected from a sensor, according to various embodiments.

[0015] [Figure 9] FIG. 1 illustrates a method for classifying contaminant source types based on sensor data, according to various embodiments.

[0016] [Figure 10] 4 illustrates a method for determining a source type of a detected first gas based on sensor data, according to various embodiments.

[0017] [Figure 11] 1 illustrates a method for providing a model that predicts leak probability at various locations within a given geographic area, according to various embodiments.

[0018] [Figure 12] 4A and 4B illustrate the relationship between detection probability distributions and leakage probability distributions in accordance with various embodiments.

[0019] [Figure 13] 1 illustrates a method for determining a leakage condition for a cluster in accordance with various embodiments.

[0020] [Figure 14] 1 illustrates a method for determining the leak status of a particular leak, according to various embodiments.

[0021] [Figure 15] 1 illustrates a method for determining the leak status of a particular leak, according to various embodiments.

[0022] [Figure 16] 10A-10C illustrate a method for associating peaks detected in sensor data with new or existing clusters associated with particular leaks, according to various embodiments.

[0023] [Figure 17] 1 illustrates a method for detecting leaks in accordance with various embodiments.

[0024] [Figure 18] 1 illustrates a method for providing leakage data in accordance with various embodiments. DETAILED DESCRIPTION OF THE INVENTION

[0025] The present invention may be embodied in various forms, including as a process, an apparatus, a system, a composition of matter, a computer program product embodied on a computer-readable storage medium, and / or a processor configured to execute instructions stored in and / or provided by a memory coupled to the processor. These embodiments, or any other form the present invention may take, may be referred to herein as technology. In general, the order of steps in a disclosed process may be varied within the scope of the present invention. Unless otherwise noted, components, such as a processor or memory, described as configured to perform a task may be implemented as general components temporarily configured to perform the task at a given time, or as specific components manufactured to perform the task. As used herein, the term “processor” refers to one or more devices, circuits, and / or processing cores configured to process data, such as computer program instructions.

[0026] The following is a detailed description of one or more embodiments of the present invention with reference to figures that illustrate the principles of the invention. While the present invention has been described in connection with such embodiments, it is not limited to any particular embodiment. The scope of the present invention is limited only by the claims, and the present invention includes many alternatives, modifications, and equivalents. In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention. These details are for the purpose of example, and the present invention may be practiced according to the claims without some or all of these specific details. For simplicity, technical matters that are well known in the art related to the present invention have not been described in detail so as not to unnecessarily obscure the present invention.

[0027] Related art systems provided an indication of leak detection, such as by generating a map for an area where the map contained an indicator of leak occurrence. However, related art systems did not intelligently shut off leaks (e.g., did not make an intelligent decision to transition a leak from an active state to an inactive state). Rather, such systems either never shut off the leak (e.g., indicated the leak as continuing long after the leak was actually repaired) or arbitrarily shut off the leak according to a predetermined time threshold after a detected leak was deemed shut off. These related art systems did not provide an accurate measure or characterization of contaminant detection. For example, related art systems did not classify the type of leak based on sensor data (e.g., did not identify the source type as pyrolytic or biogenic) and did not utilize robust techniques for determining when a leak started and when it ended. This low-resolution data does not allow organizations (e.g., regulatory agencies, local governments, public utilities) to accurately determine the extent of contamination caused by the leak.

[0028] Various embodiments provide systems, methods, and apparatus for probabilistically characterizing leaks. The system deploys a model (e.g., a hidden Markov model) to predict whether a leak is active or inactive (e.g., leak-on state or no-leak state). The model uses robust statistical analysis to provide accurate predictions of when a leak will be detected, where the leak is occurring (e.g., clustering positive contaminant detections into different leaks), and when the leak will transition to an inactive state. Accurate determination of the leak's lifetime (e.g., leak start date and leak end date) allows for more accurate determination of the total contamination associated with the leak, which can be utilized in connection with imposing fines, deploying repair resources, etc.

[0029] In some embodiments, the system clusters positive contaminant detections as leaks. The system re-runs the model for each date in the cluster's history to determine when the model predicts that the leak has stopped. Parameters for the model are adjusted to balance the competing goals of having a high degree of confidence that the leak has been shut off (e.g., transitioning to an inactive state) and not delaying the transition from an active leak to an inactive leak because the model has waited long enough for the leak to be shut off with sufficient confidence (e.g., the model has waited too many leak-free observations before determining that the leak is inactive). Based on the model parameter adjustments, the model can be set to take a long time to declare that the leak has stopped with high confidence, or at the other end of the spectrum, the model may be adjusted to be too aggressive, which could lead to the model erroneously concluding that the leak has stopped based on one or more detections of no peaks.

[0030] In some embodiments, the system determines whether a detection is pyrolytic or biogenic. In the case of methane detection, the system accurately classifies the detection as pyrolytic or biogenic based at least in part on the detection of the co-occurrence of ethane. Pyrolytic sources of methane exhibit ethane co-occurrence, while biogenic sources of methane lack a corresponding ethane emission. Using ethane in the context of leak classification is challenging when the intensity of the emitted ethane is low compared to the intensity of the methane emission during the leak. Because the intensity of the ethane detection is so low, the system may confuse the ethane detection with sensor noise. The system utilizes statistical analysis in connection with calibrating the sensor and detecting ethane emission in the face of a baseline sensor noise. By classifying methane leaks as pyrolytic or biogenic, the system can provide organizations (e.g., local governments, utilities, regulatory agencies, etc.) with an accurate indication of the methane source type (e.g., the system identifies leaks originating from gas pipelines, etc.).

[0031] Various embodiments provide systems, devices, and methods for classifying gas signals. The methods include: (i) receiving sensor data collected over a geographic region from one or more mobile sensors; (ii) detecting a first gas signal in the sensor data; (iii) determining a source type of the first gas based at least in part on a determination of whether the sensor data includes a signal of another contaminant; and (iv) providing the source type. The first gas signal may correspond to a methane gas signal. As an example, the other contaminant may correspond to ethane. In some embodiments, the source type is deemed to correspond to a biogenic source type in response to a determination that the sensor data includes a methane gas signal without the presence of an ethane gas signal. In some embodiments, the source type is deemed to correspond to a pyrolytic source type in response to a determination that the sensor data includes a methane gas signal and an ethane gas signal. As an example, the system may deem the sensor data to include an ethane signal in response to a determination that an ethane measurement in the sensor data exceeds a noise baseline by a predetermined amount. The predetermined degree corresponds to at least 300% of the noise baseline (e.g., the ethane signal is three times the noise baseline). In one example, the noise baseline is a rolling baseline over a predetermined time window.

[0032] Various embodiments provide systems, devices, and methods for ranking sub-regions within a geographic region based on the severity of a gas leak. The method includes: (i) obtaining sensor data collected across the geographic region, the sensor data being collected by one or more mobile sensors; (ii) determining a model for predicting leak probability for one or more sub-regions within the geographic region based at least in part on the sensor data, the detection probability, and the collection intensity; and (iii) providing the model. In some embodiments, the model is used in connection with determining when a leak starts and when a leak ends. As an example, the one or more sub-regions include one or more road segments within the geographic region. In some embodiments, the detection probability includes a leak component corresponding to the probability of detecting a leak and a non-leak component corresponding to the probability of detecting a non-leak. The detection probability may further include the probability of detecting a leak when no leak is present.

[0033] Various embodiments are disclosed that provide systems, devices, and methods for detecting leaks near an emission source, the method comprising: (i) receiving a first information stream indicative of a leak condition from one or more moving sensors, (ii) determining that an initial leak condition exists based at least in part on the first information stream indicative of the leak condition, (iii) receiving a second information stream indicative of a non-leak condition, (iv) using a statistical model to determine that the leak condition has ended based at least in part on the first information stream and the second information stream, (v) receiving a third information stream indicative of the leak condition, and (vi) determining that a new leak condition exists, the new leak condition being a different leak condition from the initial leak condition.

[0034] Hyper-local environmental data (e.g., data related to air quality and greenhouse gases) can be collected using installed air pollutant sensors. Embodiments of techniques that can be used to collect hyper-local data are described in commonly assigned U.S. patent application Ser. No. 16 / 682,871, filed November 13, 2019, entitled "HYPER-LOCAL MAPPING OF ENVIRONMENTAL CONDITIONS," commonly assigned U.S. patent application Ser. No. 16 / 409,624, filed May 10, 2019, entitled "INTEGRATION AND ACTIVE FLOW CONTROL FOR ENVIRONMENTAL SENSORS," commonly assigned U.S. patent application Ser. No. 16 / 773,873, filed January 27, 2020, entitled "SENSOR DATA AND PLATFORMS FOR VEHICLE ENVIRONMENTAL QUALITY MANAGEMENT" (commonly assigned U.S. patent application Ser. No. 62 / 798,395, entitled "SENSOR DATA AND PLATFORMS FOR VEHICLE ENVIRONMENTAL QUALITY MANAGEMENT"). No. 60 / 699,992, filed on Dec. 1, 2002, which claims priority under "Patent Document 1: MANAGEMENT," all of which are incorporated herein in their entirety for all purposes.

[0035] FIG. 1 illustrates one embodiment of a system 100 for collecting and processing environmental data. The system 100 includes multiple mobile sensor platforms 102A, 102B, and 102C and a server 150. In some embodiments, the system 100 may further include one or more fixed sensor platforms 103, one of which is shown. The fixed sensor platform 103 may be used to collect environmental data at a fixed location. The environmental data collected by the fixed sensor platform 103 may supplement the data collected by the mobile sensor platforms 102A, 102B, and 102C. Thus, the fixed sensor platform 103 may have the same or similar sensors as the mobile sensor platforms 102A, 102B, and 102C. In another embodiment, the fixed sensor platform 103 may be omitted. Although one server 150 is shown, multiple servers may be present. The multiple servers may be located in different locations. Although three mobile sensor platforms 102A, 102B, and 102C are shown, there are typically other numbers of sensors / mobile sensor platforms. The mobile sensor platforms 102A, 102B, and 102C and the fixed sensor platform 103 may communicate with the server 150 via a data network 108. The communication may be wireless.

[0036] The mobile sensor platforms 102A, 102B, and 102C may be mounted on vehicles (such as automobiles or drones). In some embodiments, it is desirable for the mobile sensor platforms 102A, 102B, and 102C to be close to the ground so that they can better sense conditions similar to those experienced by humans. The mobile sensor platform 102A includes a bus 106 and sensors 110, 120, and 130. Although three sensors are shown, a different number of sensors may be present on the mobile sensor platform 102A. Furthermore, other configurations of components may be used with the sensors 110, 120, and 130. Each sensor 110, 120, and 130 is used to sense an environmental quality and may be of primary interest to a user of the system 100. For example, sensors 110, 120, and 130 may be gas sensors, volatile organic compound (VOC) sensors, particulate matter sensors, radiation sensors, noise sensors, light sensors, temperature sensors, sound sensors, or other similar sensors that capture environmental fluctuations. For example, sensors 110, 120, and 130 may be used to sense one or more of NO2, CO, NO, O3, SO2, CO2, VOCs, CH4, particulate matter, noise, light, temperature, radiation, and other compounds. In some embodiments, sensors 110, 120, and / or 130 may be multi-modality sensors. Multi-modality gas sensors sense multiple gases or compounds. For example, if sensor 110 is a multi-modality NO2 / O3 sensor, sensor 110 can sense both NO2 and O3 simultaneously. Sensor 110 may include multiple sensors (such as sensors 112, 114, and 116). Sensor 120 may include multiple sensors (such as sensors 122, 124, and 126). Sensor 130 may include multiple sensors (such as sensors 132 and 134).

[0037] Although not shown in FIG. 1 , other sensors co-located with sensors 110, 120, and 130 may be used to sense characteristics of the surrounding environment (such as other gases and / or substances in some examples). Such additional sensors are exposed to the same environment as sensors 110, 120, and 130. In some embodiments, such additional sensors are in close proximity (e.g., within 10 millimeters or less) to sensors 110, 120, and 130. In some embodiments, the additional sensors may be remote from sensors 110, 120, and 130 if they sample air within the same space within a closed system (such as a closed-tube system). In some embodiments, temperature and / or pressure are sensed by these additional sensors. For example, the additional sensors co-located with sensor 110 may be temperature, pressure, and relative humidity (T / P / RH) sensors. These additional co-located sensors may be used to calibrate sensors 110, 120, and / or 130. Although not shown, the sensor platform 102A may further include a manifold for drawing in and delivering air to the sensors 110, 120, and 130 for testing.

[0038] The sensors 110, 120, and 130 provide sensor data over the bus 106 or via another mechanism. In some embodiments, the data from the sensors 110, 120, and 130 includes a time. This time may be provided by a master clock (not shown) and may take the form of a timestamp. The master clock may reside on the sensor platform 102A, be part of the processing unit 140, or be provided by the server 150. As a result, the sensors 110, 120, and 130 may provide time-stamped sensor data to the server 150. In other embodiments, the time associated with the sensor data may be provided in other ways. Although the sensors 110, 120, and 130 typically acquire data at a particular frequency, and therefore the sensor data is discussed as being associated with a particular time interval (e.g., a period related to the frequency), the sensor data may be time-stamped with a particular value. For example, sensors 110, 120, and / or 130 may acquire sensor data every 1 second, 2 seconds, 10 seconds, or 30 seconds. The time interval may be 1 second, 2 seconds, 10 seconds, or 30 seconds. The time interval may be the same for all sensors 110, 120, and 130, or may be different for different sensors 110, 120, and 130. In some embodiments, the time interval of a sensor data point is centered around a timestamp. For example, if the time interval is 1 second and the timestamp is t1, the time interval may be from t1-0.5 seconds to t1+0.5 seconds. However, other mechanisms for defining the time interval may be utilized.

[0039] The sensor platform 102A further includes a position unit 145 that provides position data. In some embodiments, the position unit 145 is a global positioning satellite (GPS) unit. Accordingly, the system 100 will be described in the context of the position unit 145. The position data may be time-stamped in a similar manner to the sensor data. Because the position data is associated with the sensor data, it may also be considered to be associated with a time interval, as described above. However, in some embodiments, the position data (e.g., GPS data) may be obtained more or less frequently than the sensor data. For example, the position unit 145 may capture position data every second, while the sensor 130 may capture data every 30 seconds. Thus, multiple data points for the position data may be associated with a single 30-second time interval. The position data may be processed as described below.

[0040] An optional processing unit 140 may perform some processing and functions on the data from the sensor platform 104, may simply pass the data from the sensor platform 104 to a server, or may be omitted.

[0041] Mobile sensor platforms 102B and 102C are similar to mobile sensor platform 102A. In some embodiments, mobile sensor platforms 102B and 102C have the same components as mobile sensor platform 102A, although in other embodiments the components may be different. However, mobile sensor platforms 102A, 102B, and 102C function similarly.

[0042] Server 150 includes a sensor data database 156, a calibration table 154 (e.g., stored in database 152), a processor 158, and memory 159. One or more processors 158 may include multiple cores. One or more processors 158 may include one or more central processing units (CPUs), one or more graphics processing units (GPUs), and / or one or more other processing devices. Memory 159 may include a first primary storage area (typically random access memory (RAM)) and a second primary storage area (typically non-volatile storage such as a solid-state drive (SSD) or hard disk drive (HDD)). Memory 159 stores programming instructions and data for processes running on processor 158. The primary storage typically includes basic operating instructions, program code, data, and objects utilized by processor 158 to perform their functions. Primary storage (e.g., memory 159) may include any suitable computer-readable storage medium described below, depending, for example, on whether data access needs to be bidirectional or unidirectional.

[0043] The sensor data database 156 includes data received from the mobile sensor platforms 102A, 102B, and / or 102C. The sensor data captured by the mobile sensor platforms 102A, 102B, and / or 102C and stored in the sensor data database 156 may be subjected to various analyses, as described below. The location data database 152 stores location data received from the mobile sensor platforms 102A, 102B, and / or 102C. In some embodiments, the sensor data database 156 stores location data and sensor data. In such embodiments, the location data database 152 may be omitted. The server 150 may include other databases and / or store and utilize other data. For example, the server 150 may include calibration data (not shown) used to calibrate the sensors 110, 120, and 130.

[0044] The system 100 may be used to acquire, analyze, and provide information about hyper-local environmental data. The mobile sensor platforms 102A, 102B, and 102C may be used to traverse a route and provide sensor data and location data to the server 150. The server 150 may process the sensor data and location data. The server 150 may also assign the sensor data to map features that correspond to the locations of the mobile sensor platforms 102A, 102B, and 102C within the same time interval in which the sensor data was captured. As described above, these map features may be hyper-local (e.g., road segments of 100 meters or less or road segments of 30 meters or less). Thus, the mobile sensor platforms 102A, 102B, and 102C may provide sensor data that can capture variations at this hyper-local distance scale. The server 150 may provide the environmental data, a score, a confidence score, and / or other evaluation of the environmental data to a user. Thus, using system 100, hyper-local environmental data may be acquired using a relatively sparse network of mobile sensor platforms 102A, 102B, and 102C, associated with hyper-local map features, and processed for improved user understanding.

[0045] 2 illustrates an example embodiment of a method 200 for acquiring environmental data using mobile sensor platforms (e.g., mobile sensor platforms 102A, 102B, and 102C). Method 200 is described in the context of system 100, but may be performed using other systems. For simplicity, only certain portions of method 200 are illustrated. Although shown in sequence, in some embodiments, operations may be performed in parallel and / or in a different order.

[0046] In step 202, a mobile sensor platform traverses a route in a geographic region. While traversing the route, the mobile sensor platform collects not only sensor data but also location data. For example, the mobile sensor platform may sense one or more of NO2, CO, NO, O3, SO2, CO2, CH4, VOCs, particulate matter, other compounds, radiation, noise, light, and other environmental data at various times during the traversal of the route. Other environmental characteristics (including, but not limited to, temperature, pressure, and / or humidity) may also be sensed in step 202. Additionally, a time corresponding to the environmental data is also obtained. The time may be in the form of a timestamp for the sensor data (sensor timestamp), and the timestamp may correspond to a particular time interval. Different sensors on the mobile sensor platform may acquire environmental data at different times and / or with different frequencies. Also in step 202, the mobile sensor platform acquires location data, for example, via a GPS unit. The location data may include the position (as indicated by a GPS unit), velocity, and / or other information related to the geographic location of the mobile sensor platform. In some embodiments, location data from other sources (such as acceleration) may be obtained from the vehicle or another source. The location data may include a timestamp (location timestamp) or other indicator of the time the location data was obtained.

[0047] The mobile sensor platforms provide location data and sensor data to the server in step 204. In some embodiments, the mobile sensor platforms provide this data substantially in real time as the mobile sensor platforms traverse their route in step 202. Thus, the location data and sensor data may be transmitted wirelessly to the server. In some embodiments, some or all of the location data and / or sensor data is stored on the mobile sensor platform and then provided to the server. For example, the data may be transferred to the server when the mobile sensor platform returns to its base. In some embodiments, the mobile sensor platform may process the sensor data and / or location data before transmitting the sensor data and / or location data to the server. In other embodiments, the mobile sensor platform provides little or no processing. The sensor data and location data may be transmitted simultaneously or separately.

[0048] In step 206, the route travel and data collection in step 202 and the data transmission in step 204 are repeated. Thus, the mobile sensor platform may travel the same or different routes in step 206. In either case, multiple passes of the same geographic location, and therefore multiple passes of the same corresponding map feature, are made in step 206. In some embodiments, the repetition in step 206 may be periodic (e.g., approximately weekly, monthly, or other period). In some embodiments, the repetition in step 206 may be performed based on other timing. In some cases, the same mobile sensor platform is sent out on the same route and / or collects data for the same map feature. In some embodiments, different mobile sensor platforms may be used for the same route and / or map feature. Also, in step 206, steps 202 and 204 may be performed multiple times. Thus, in step 206, data for a particular area may be aggregated over time.

[0049] For example, FIGS. 3A-3C illustrate a particular geographic region and a route that may be traversed using method 200. A map 300 of the geographic region is shown in FIG. 3A. Map 300 may be an open-source map or may be generated by another mapping tool. Map 300 includes streets 310 (vertically on the page) and 312 (horizontally on the page), a wider street / highway 314, structures 320 and 322, and an open area 324. For simplicity, only one of each structure 320 and 322 is labeled. Open area 324 may correspond to a park, vacant lot, or similar item. As seen in FIG. 3A, the density and size of structures 320 and 322 vary across map 300. Similarly, the density and size of streets 312, 314, and 320 vary. Furthermore, structure 322 is more clearly separated by open areas, which may correspond to gardens or similar areas.

[0050] FIG. 3B illustrates a map 300 and a route 330 that may be traversed by a mobile sensor platform (such as mobile sensor platform 102A). In step 202, mobile sensor platform 102A may traverse route 330. As seen in FIG. 3B, route 330 includes portions of each street 312 and 314 in map 300. Some portions of some streets are traversed multiple times along the same route 330. In some embodiments, this still counts as a single traversal of those streets. As mobile sensor platform 102A traverses route 330 in step 202, sensor data is acquired by sensors 110, 120, and 130. Also in step 202, location data is acquired by location unit 145 throughout route 330. In some embodiments, the vehicle carrying mobile sensor platform 102A moves slowly enough while traversing route 330 so that sensor and location data can be accurately acquired for a particular location. In some embodiments, the mobile sensor platform 102A moves at a speed that allows for multiple sensor data points for each map feature. The mobile sensor platform 102A also transmits location data and sensor data to the server 150 in step 204. This may be done while the mobile sensor platform 102A traverses the route 330 or afterward. Other mobile sensor platforms 102B and / or 102C may also traverse the same or different routes and transmit data to the server 150 in steps 202 and 204. Thus, multiple mobile sensor platforms may be used in the method 200.

[0051] In step 206, the mobile sensor platform 102A and / or the other mobile sensor platforms 102B and 102C repeat the route traversal, data collection, and transmission of location and sensor data. In some cases, the mobile sensor platforms 102A, 102B, and / or 102C traverse the route 330 again. In some cases, the mobile sensor platforms 102A, 102B, and / or 102C traverse another route. For example, FIG. 3C shows a map 300 with another route 332. As part of step 206, the mobile sensor platforms 102A, 102B, and / or 102C may traverse the route 332 and collect location and sensor data in step 206 (repeat step 202). In some embodiments, the vehicle carrying the mobile sensor platform 102A, 102B, and / or 102C moves slowly enough while traversing the route 332 so that sensor and location data can be accurately acquired for a particular location. In some embodiments, the mobile sensor platform 102A, 102B, and / or 102C moves at a speed that allows for multiple sensor data points for each map feature (described below). During or after traversing the route 330 and / or route 332, the mobile sensor platform 102A, 102B, and / or 102C transmits the sensor and location data to the server 150 in step 206 (repeat step 204).

[0052] Thus, using method 200, sensor data and location data may be obtained for an area of ​​a map. The sensor data and location data may be provided to server 150 or other components for processing, aggregation, and analysis. The sensor data and location data are sensed using method 200 frequently enough so that variations in environmental quality at a hyperlocal scale can be reflected in the sensor data. Method 200 may be performed using a relatively small number of mobile sensor platforms. As a result, efficiency of data collection may be improved while maintaining sufficient sensitivity for both the sensor data and the location data.

[0053] For example, FIGS. 3A-3C illustrate a particular geographic region and a route that may be traversed using method 200. A map 300 of the geographic region is shown in FIG. 3A. Map 300 may be an open-source map or may be generated by another mapping tool. Map 300 includes streets 310 (vertically on the page) and 312 (horizontally on the page), a wider street / highway 314, structures 320 and 322, and an open area 324. For simplicity, only one of each structure 320 and 322 is labeled. Open area 324 may correspond to a park, vacant lot, or similar item. As seen in FIG. 3A, the density and size of structures 320 and 322 vary across map 300. Similarly, the density and size of streets 312, 314, and 320 vary. Furthermore, structure 322 is more clearly separated by open areas, which may correspond to gardens or similar areas.

[0054] FIG. 3B illustrates a map 300 and a route 330 that may be traversed by a mobile sensor platform (such as mobile sensor platform 102A). In step 202, mobile sensor platform 102A may traverse route 330. As seen in FIG. 3B, route 330 includes portions of each street 312 and 314 in map 300. Some portions of some streets are traversed multiple times along the same route 330. In some embodiments, this still counts as a single traversal of those streets. As mobile sensor platform 102A traverses route 330 in step 202, sensor data is acquired by sensors 110, 120, and 130. Also in step 202, location data is acquired by location unit 145 throughout route 330. In some embodiments, the vehicle carrying mobile sensor platform 102A moves slowly enough while traversing route 330 so that sensor and location data can be accurately acquired for a particular location. In some embodiments, the mobile sensor platform 102A moves at a speed that allows for multiple sensor data points for each map feature. The mobile sensor platform 102A also transmits location data and sensor data to the server 150 in step 204. This may be done while the mobile sensor platform 102A traverses the route 330 or afterward. Other mobile sensor platforms 102B and / or 102C may also traverse the same or different routes and transmit data to the server 150 in steps 202 and 204. Thus, multiple mobile sensor platforms may be used in the method 200.

[0055] In step 206, the mobile sensor platform 102A and / or the other mobile sensor platforms 102B and 102C repeat the route traversal, data collection, and transmission of location and sensor data. In some cases, the mobile sensor platforms 102A, 102B, and / or 102C traverse the route 330 again. In some cases, the mobile sensor platforms 102A, 102B, and / or 102C traverse another route. For example, FIG. 3C shows a map 300 with another route 332. As part of step 206, the mobile sensor platforms 102A, 102B, and / or 102C may traverse the route 332 and collect location and sensor data in step 206 (repeat step 202). In some embodiments, the vehicle carrying the mobile sensor platform 102A, 102B, and / or 102C moves slowly enough while traversing the route 332 so that sensor and location data can be accurately acquired for a particular location. In some embodiments, the mobile sensor platform 102A, 102B, and / or 102C moves at a speed that allows for multiple sensor data points for each map feature (described below). During or after traversing the route 330 and / or route 332, the mobile sensor platform 102A, 102B, and / or 102C transmits the sensor and location data to the server 150 in step 206 (repeat step 204).

[0056] Thus, using method 200, sensor data and location data may be obtained for an area of ​​a map. The sensor data and location data may be provided to server 150 or other components for processing, aggregation, and analysis. The sensor data and location data are sensed using method 200 frequently enough so that variations in environmental quality at a hyperlocal scale can be reflected in the sensor data. Method 200 may be performed using a relatively small number of mobile sensor platforms. As a result, efficiency of data collection may be improved while maintaining sufficient sensitivity for both the sensor data and the location data.

[0057] In some embodiments, the system accurately classifies leaks based on contaminant signals detected above the sensor noise baseline. Particular contaminants can originate from different types of sources. For example, methane emissions can be attributed to pyrolytic sources (e.g., fossil fuel-derived sources) or biogenic sources (e.g., biological sources such as bacteria). The system processes the sensor data to detect contaminant signals in the sensor data in order to properly classify the contaminant (e.g., properly attribute the contaminant to a particular source type). For example, a utility customer may only be interested in leaks over a geographic area that originate from pyrolytic sources because they can redirect maintenance resources to identified leaks to the location where the leak was detected. Without accurate source type classification, the utility customer may inappropriately deploy maintenance resources to sources that they cannot control (e.g., biogenic sources such as plant-animal material, landfill releases, wetlands, etc.).

[0058] In some embodiments, the system estimates the source type of detected methane based on whether the sensor data includes the co-occurrence of ethane. Sources of pyrogenic methane (e.g., typically natural gas leaks) are distinguished from biogenic sources based on the co-occurrence of a certain percentage of high ethane levels. In contrast, biogenic methane occurs alone because it is produced by bacteria that produce only methane without ethane.

[0059] According to various embodiments, despite the unstable baseline, the acquired ethane signal exhibits hyperlocal enhancements. Various embodiments implement a baseline subtraction model configured to capture only the exhibited hyperlocal enhancements. Instead of raw ethane concentrations, the baseline-subtracted enhancements may be treated as "modalities" that are propagated through aggregation processes (e.g., pass-through, segment median, etc.) and categorized into "low," "medium," or "high" classifications of contaminant strength.

[0060] FIG. 4 illustrates the relationship between methane and ethane signals in collected sensor data according to various embodiments. Graph 400 illustrates the average level / change of methane versus the average level / change of ethane. For each sampling at a particular location, the sensor data includes a correlation between ethane and methane buildup (e.g., ethane and methane signals). The system may discard signals that occur with buildup of carbon monoxide based on the assumption that such signals are due to vehicular sources (e.g., not leaks in gas-carrying infrastructure). As shown, the amount of ethane present in the sensor data for sources of pyrolysis origin is quite small. While methane levels may have intensities on the order of thousands, ethane levels have intensities on the order of hundreds. Due to minimal ethane detection, the system may incorrectly classify a source as of pyrolysis origin based on detection of an ethane signal due to sensor noise.

[0061] In various embodiments, the system determines the sensor noise inherent in a sensor (e.g., a set of mobile sensors performing sampling over a geographic region). The system accurately classifies the contaminant source type despite the significant contribution of sensor noise in the collected sensor data. For example, the system adjusts for the probability of detection or the likelihood of a false positive.

[0062] FIG. 5A illustrates an example of the calibration of a set of sensors, according to various embodiments. In the illustrated example, a calibration graph 500 shows the noise observed across a set of sensors over different time periods. Each bar represents a different sensor, with the same sensor analyzed for different calibration periods. For example, the leftmost bar in the graph of data observed during the low calibration period is the same sensor as the leftmost bar in the graph of data observed during the high calibration period. The calibration of the sensors shown in FIG. 5A may correspond to a particular contaminant (e.g., ethane). During both the low and medium calibration periods, the observed noise is approximately 2.5 parts per billion. In some embodiments, the system determines a statistic (e.g., mean, median, etc.) of the noise observed across the set of sensors and considers that statistic to be the sensor noise baseline.

[0063] In some embodiments, the system calibrates sensors by exposing a set of sensors to a particular contaminant at a particular contaminant strength, taking sensor data collected by the set of sensors, and controlling for the contribution of known contaminants to derive sensor noise.

[0064] The system accurately classifies contaminant sources by detecting statistically relevant contaminant signals while taking into account the sensor baseline. In some embodiments, the system classifies contaminant sources based on utilizing contaminant signals having intensities greater than the sensor noise baseline by a predetermined absolute amount. In some embodiments, the system classifies contaminant sources based on utilizing contaminant signals having intensities greater than the sensor noise baseline by a predetermined degree. One example of a predetermined degree used to identify contaminant contributions not attributable to sensor noise is three times the sensor noise baseline. Various other predetermined degrees may also be implemented (e.g., two times the sensor noise baseline, 2.5 times the sensor noise, etc.).

[0065] In some embodiments, the system uses a moving sensor noise baseline (such as a moving median baseline). The moving median sensor noise baseline may be adjusted by adjusting the window (e.g., time period) over which the median sensor noise baseline is calculated. The system may adjust the window based on the goodness of fit between the moving median sensor noise baseline and the signal contained in the sensor data.

[0066] 5B and 5C show examples of moving baseline signals according to a first time window for methane and ethane signals, according to various embodiments. As shown, the moving median sensor noise baseline 525 for methane, shown in graph 520, and the moving median sensor noise baseline 545 for ethane, shown in graph 540, are not a good fit to the signals. As an example, the window used to calculate the moving median sensor noise baseline 525 and the moving median sensor noise baseline 545 may be 20 minutes. The system may adjust the window to obtain a better fit between the sensor noise baseline and the signal. For example, the system may adjust the window over which the moving median sensor noise baseline is calculated to 5 minutes.

[0067] 5D and 5E show example moving baseline signals according to a second time window for methane and ethane signals, according to various embodiments. As shown, the moving median sensor noise baseline 555 for methane, shown in graph 550, and the moving median sensor noise baseline 565 for ethane, shown in graph 560, fit the signals more closely compared to graphs 520 and 540 in FIGS. 5B and 5C. In other words, adjusting the window, such as by shortening the window from 20 minutes to 5 minutes, results in a baseline that better captures nearby scale variations in the signal.

[0068] 5F illustrates an example of an ethane signal relative to various baselines, according to various embodiments. In the illustrated example, graph 580 shows a signal 582 collected with sensor data and a running median sensor noise baseline 584 closely fitted to signal 582. Illustrated are example contaminant thresholds 586, 588 that can be used to detect contaminant contributions to the signal. For example, the system uses a contaminant threshold that is statistically relevant or significantly higher than the running median sensor noise baseline to avoid false positives.

[0069] In response to receiving the sensor data, the system extracts the contaminant signal by separating the contribution of the contaminant from the sensor noise, such as by counting measurements having a contaminant intensity greater than a contaminant threshold relative to a running median sensor noise baseline as a contaminant detection. In some embodiments, the system classifies the source type of a first contaminant signal (e.g., a methane signal) based at least in part on an observed second contaminant signal (e.g., an ethane signal), if any. For example, the system determines that ethane was detected as co-occurring with methane based at least in part on determining that the observed ethane measurement has an intensity greater than a contaminant threshold (e.g., three times the sensor noise baseline). In response to determining that a second contaminant (e.g., ethane) co-occurs with the first contaminant (e.g., methane), the system determines that the source type of the first contaminant source is pyrolytic in origin. Conversely, if the first contaminant is observed without a statistically significant observation of the second contaminant, the system determines that the source type of the first contaminant source is biogenic in origin.

[0070] In some embodiments, in response to classifying pollutant source types, the system can map observed leaks or pollutants emitted from biogenic sources across a geographic area. Utility customers can use the leak map to deploy maintenance resources. Local governments or regulatory agencies can use the map of pollutants emitted from biogenic sources in connection with determining the impact of biogenic sources or otherwise modifying areas to mitigate biogenic contributions to the atmosphere.

[0071] 6A and 6B show example maps including methane signals from different source types, according to various embodiments. In the illustrated example, map 600 shows a leak (e.g., methane observed in conjunction with an ethane signal). Map 650 shows observations of methane originating from a biogenic source. Highlighted road segments correspond to the observations.

[0072] In some embodiments, if the pollutant intensity (e.g., methane increase) is too small (e.g., below a predetermined threshold or deemed statistically irrelevant) for the system to identify a methane detection, the system does not classify the methane detection. The system can use the actual number of days a pollutant is detected at a particular location as a good measure to characterize the detection / leak, such as in connection with determining whether to classify a pollutant. For example, the actual number of days the system observes a sufficient amount of pollutant to be classified can be used as an indicator of source persistence. The system can further characterize a detected leak based on the hit rate of pollutant detection at a particular location (e.g., a road segment-level hit rate). The system calculates the road segment-level hit rate as the number of detections of a pollutant with an intensity large enough to be classified divided by the total number of samples made on that segment (e.g., the number of passes by vehicles carrying one or more mobile sensors). The hit rate can be used as an indicator of source persistence and / or detectability.

[0073] In some embodiments, the system represents the detections on a map. For example, the system generates a map including a representation of the detections, which may be labeled (e.g., color-coded) according to the actual number of days the system observed sufficient quantities of contaminants to be classified. As another example, the system generates a map including a representation of the detections, which may be labeled (e.g., color-coded according to hit rate). Hit rates may be characterized as "low," "medium," or "high," each with a corresponding range of hit rates.

[0074] 7 illustrates a method for calibrating a sensor, according to various embodiments. In some embodiments, process 700 is performed at least in part by system 100 of FIG. 1. Process 700 may be performed in connection with calibrating a sensor (e.g., a sensor of a particular sensor type).

[0075] In step 705, the system receives an instruction to determine the sensor noise baseline.

[0076] In step 710, a sensor is selected.

[0077] In step 715, the system exposes the selected sensor to a predetermined intensity of a predetermined contaminant. In some embodiments, the predetermined contaminant is methane. Various other contaminants may be used. The contaminant may be selected at least in part based on the contaminants for which the system detects and classifies a leak condition.

[0078] At step 720, the system acquires sensor data from the selected sensors. For example, the system acquires sensor data generated while the sensors were collecting measurements when exposed to a predetermined contaminant. At step 725, the system determines whether a sensor noise baseline is to be determined for additional sensors. In response to determining that a sensor noise baseline is to be determined for additional sensors, process 700 returns to step 710, where process 700 repeats steps 710-720 until the system determines that no additional sensor noise baselines remain to be determined. In response to determining that no additional sensor noise baselines remain to be determined, process 700 proceeds to step 730. At step 730, the system determines a sensor noise baseline based at least in part on the sensor data. In some embodiments, the system determines the sensor noise baseline based on a statistical analysis of sensor data acquired from a set of sensors (e.g., sensors of the same type). For example, the sensor noise baseline is considered to be the average sensor noise across the set of sensors. As another example, the sensor noise baseline is considered to be the median sensor noise across the set of sensors. As another example, a sensor noise baseline is a statistically relevant measure of noise predicted to be inherent in a sensor based on a set of sensor behaviors.

[0079] In step 735, the system provides an indication of the sensor noise baseline. The system may provide an indication of the sensor noise baseline to another system or service (such as the system or service that invoked process 700). As an example, the system provides the sensor noise baseline to a system / service configured to perform leak detection, leak classification, leak condition prediction, etc.

[0080] At step 740, a determination is made as to whether process 700 is complete. In some embodiments, process 700 is determined to be complete in response to a determination that there are no more sensors to be analyzed, no more sensor noise baselines to be determined, no more sensor analyses to be performed for other contaminants, an administrator indicating that process 700 is to be paused or stopped, etc. In response to a determination that process 700 is complete, process 700 ends. In response to a determination that process 700 is not complete, process 700 returns to step 705.

[0081] 8 illustrates a method for determining the acquisition of different contaminant signals in collected data collected from a sensor, according to various embodiments. In some embodiments, process 800 is performed at least in part by system 100 of FIG.

[0082] In step 805, the system receives an instruction to analyze the sensor data.

[0083] In step 810, the system acquires sensor data. The sensor data may be collected by one or more mobile sensors (e.g., a sensor platform mounted on a vehicle). For example, the sensor data corresponds to a particular sampling, such as a travel date, when a vehicle traveled a corresponding road segment and performed the sampling (e.g., collected air quality / pollutant measurements).

[0084] In step 815, the system obtains a first contaminant signal in the sensor data. The system may obtain the first contaminant signal from the suggestion provided by step 735 of process 700. Additionally or alternatively, the system may analyze the sensor data to extract the first contaminant signal contained in the sensor data (e.g., determine the contribution of the first contaminant in).

[0085] In step 820, the system obtains the second contaminant signal in the sensor data. The system may obtain the first contaminant signal from the suggestion provided by step 735 of process 700.

[0086] In step 825, the system provides an indication of the first pollutant signal and the second pollutant signal. In some embodiments, the system provides the indication of the first pollutant signal and the second pollutant signal to a system or service configured to classify a source type associated with a particular pollutant (e.g., the first pollutant).

[0087] At step 830, a determination is made as to whether process 800 is complete. In some embodiments, process 800 is determined to be complete in response to a determination that there are no more leak data to be provided, no more air quality measurements to be collected, no more leaks to be analyzed, no more pollutant signals to be detected, no more pollutant signals to be provided, an administrator indicating that process 800 is to be paused or stopped, etc. In response to a determination that process 800 is complete, process 800 ends. In response to a determination that process 800 is not complete, process 800 returns to step 805.

[0088] FIG. 9 illustrates a method for classifying the source type of a pollutant based on sensor data, according to various embodiments. In some embodiments, process 900 is performed, at least in part, by system 100 of FIG. 1. The system may infer the source type based on the presence / co-occurrence of certain pollutants. For example, if the system is classifying the source of methane, the system analyzes whether ethane is co-occurring (e.g., at a sufficient level above the sensor noise baseline). Based on the co-occurrence of high ethane levels (e.g., a certain percentage), the system infers that the source type is thermogenic, i.e., resulting from the transportation / utilization of fossil fuels. Conversely, because biogenic methane is solely produced by bacteria or other living matter that produce ethane without the co-occurrence of ethane, the system infers that the source type is biogenic in the absence of any / sufficient ethane.

[0089] In step 905, the system receives an instruction to classify the sensor data. For example, the system determines to analyze the sensor data to detect pollution signals and identify the source type of the pollutant.

[0090] At step 910, the system acquires sensor data. Additionally or alternatively, the system receives an indication of the first contaminant signal and / or the second contaminant signal in the sensor data. For example, the system may receive the indication from a system or service at step 825 of process 800.

[0091] At step 915, the system determines whether the sensor data includes a first contaminant signal. The first contaminant signal may correspond to a measurement collected for a particular contaminant (e.g., the contaminant being monitored). In some embodiments, the first contaminant is methane. In response to determining that the sensor data does not include the first contaminant signal, process 900 proceeds to step 945. Conversely, in response to determining that the sensor data does include the first contaminant (e.g., methane), process 900 proceeds to step 920.

[0092] At step 920, the system determines whether the sensor data includes a second contaminant signal. The second contaminant signal may correspond to measurements collected for another particular contaminant (such as another contaminant correlated with the first contaminant). In response to determining that the sensor data does not include the second contaminant signal, process 900 proceeds to step 930. Conversely, in response to determining that the sensor data does include the second contaminant signal, process 900 proceeds to step 925.

[0093] At step 925, the system determines whether the second contaminant signal exceeds the sensor noise baseline by a predetermined amount. The system may obtain the sensor noise baseline from step 735 of process 700. The system may determine the sensor noise baseline for a particular type of sensor. The sensor noise baseline may be updated periodically (e.g., to detect drift) or in response to finding anomalies in measurements collected from the sensor data. In some embodiments, the sensor noise baseline may be a running median of the sensor noise. The time window over which the running median is calculated may be 5 minutes. In some embodiments, the time window over which the running median is calculated is shorter than 20 minutes. Various other time windows may be used, and the configuration of the time window may be used to adjust the sensitivity of the model to identify a baseline that better fits the signals in the sensor data.

[0094] In some embodiments, the predetermined degree is a threshold value relative to the noise sensor baseline. As one example, the predetermined degree may be three times the sensor noise baseline. As another example, the predetermined degree may be two times the sensor noise baseline. However, various other relative values ​​may be implemented. Alternatively, the predetermined degree may correspond to an absolute value greater than the sensor noise baseline.

[0095] In response to determining that the second contaminant signal in the sensor data does not exceed the sensor noise baseline by a predetermined amount, process 900 proceeds to step 930, where the system determines that the source type of the first contaminant is a first source type. For example, in response to determining that the measured amount of the second contaminant is absent or below a predetermined threshold / amount, the system considers the source of the first contaminant to be a particular source type. In some embodiments, the first contaminant is methane and the second contaminant is ethane. In response to detecting methane in the absence of any ethane or sufficient levels of ethane, the system determines that the source type of the first contaminant is a biogenic source type (e.g., arising from biological material).

[0096] In response to determining that the second contaminant signal in the sensor data exceeds the sensor noise baseline by a predetermined amount (e.g., a certain percentage), process 900 proceeds to step 935, where the system determines that the source type of the first contaminant is the second source type. For example, in response to detecting sufficient co-occurrence of the first contaminant and the second contaminant, the system infers that the source type is the second source type. In some embodiments, the first contaminant is methane and the second contaminant is ethane. In response to determining that ethane sufficiently co-occurs with the methane, the system infers that the source type is a pyrolytic source type (e.g., where the methane is derived from a fossil fuel).

[0097] In response to determining the source type, process 900 proceeds to step 940. At step 940, the system provides a suggestion of the sensor data classification. For example, the system classifies the source type of the first contaminant and provides the classification. The suggestion may be provided to another system or service. For example, the system may decide to provide the suggestion to a mapping service that maps leaks across the geographic area being monitored.

[0098] At step 945, a determination is made as to whether process 900 is complete. In some embodiments, process 900 is determined to be complete in response to a determination that there are no more leak data to be provided, no more air quality measurements to be collected, no more leaks to be analyzed, no more source types of a particular pollutant to be classified, an administrator indicating that process 900 is to be paused or stopped, etc. In response to a determination that process 900 is complete, process 900 ends. In response to a determination that process 900 is not complete, process 900 returns to step 905.

[0099] 10 illustrates a method for determining a source type of a detected first gas based on sensor data, according to various embodiments. In some embodiments, process 1000 is performed at least in part by system 100 of FIG.

[0100] At step 1005, the system receives collected sensor data for a geographic region. The geographic region may be a predetermined area where air quality (e.g., pollutant levels) is monitored, such as a contract area with a particular customer (e.g., a municipality, a utility, etc.). At step 1010, the system detects a first gas signal in the sensor data. The first gas signal may be methane. In some embodiments, detecting the first gas signal includes determining that an intensity of methane in the sensor data is greater than a predetermined threshold and / or persists for a predetermined period of time. In some embodiments, the system determines that the first gas signal is detected in the sensor data in response to determining that the sensor data indicates a leak. At step 1015, the system determines a source type of the first gas based at least in part on determining whether the sensor data includes a signal for another predetermined pollutant. In some embodiments, the system classifies the source of the first gas (e.g., methane) as pyrolytic or fossil-fuel derived based on determining that the first gas co-occurs with another pollutant (e.g., ethane). The system provides the source types to another system or service. For example, a mapping service monitoring leaks from sources of pyrolytic origin may use the source type classification to appropriately identify corresponding leaks on a map across a geographic area. As another example, a mapping service may map leaks for different source types and provide different labeling or indicators of the source types. At step 1020, a determination is made as to whether process 1000 is complete. In some embodiments, process 1000 is determined to be complete in response to a determination that there is no more leak data to be provided, no more air quality measurements to be collected, no more leaks to be analyzed, no more source types of a particular pollutant to be classified, an administrator indicating that process 1000 is paused or stopped, etc. In response to a determination that process 1000 is complete, process 1000 terminates. In response to a determination that process 1000 is not complete, process 1000 returns to step 1005.

[0101] In some embodiments, a set of mobile sensors (e.g., mobile sensors mounted on a fleet of one or more vehicles) is deployed for an air quality collection session. The set of mobile sensors is deployed to measure air quality (e.g., detect pollutant levels, etc.) over a geographic area (e.g., a contract area). The set of mobile sensors generally does not sample everywhere at all times during each session. Furthermore, during each sampling (e.g., while driving on a road segment and collecting air quality measurements), leaks are not detected 100 percent of the time. For example, leaks are generally detected (e.g., detection probability) approximately 50-60 percent of the times that the location is sampled. Furthermore, some leak detections are significantly lower, such as 50 percent of the time. The leak detection probability may vary at different locations based on characteristics of the particular location, such as geography, topography, wind profile, time of day (e.g., a leak signal may experience interference during rush hour when a significant number of vehicles are emitting pollutants), underground structures, infrastructure (e.g., the type of asset / pipeline used to transport a particular gas), etc.

[0102] In some embodiments, the system generates a model for each leak (e.g., a detected leak). The model includes setting the state of the leak at different times based on available information. Assuming that the leak detection probability is less than 100 percent and that each leak has a characteristic frequency, the system uses a statistical / probabilistic model to evaluate the leak state (e.g., to determine whether to update the state to inactive, etc.).

[0103] The system may use a predetermined detection probability (e.g., a probability determined based on historical sensor data) or an average detection probability of all leaks (e.g., all leaks in a geographic region at all times for which sensor data is available, all current leaks, all leaks in a geographic region for a predetermined period of time, etc.). However, using a predetermined or average detection probability can make it easy to erroneously turn off a leak (e.g., set a leak as inactive) or turn a leak on. For example, if a 15% average detection probability is implemented and a leak is detected in the sensor data a much higher percentage of times than the average detection probability (e.g., the leak is detected with a much higher probability, or the leak is detected with a much lower probability), the model for the leak (e.g., a model for setting the state of the leak) will be surprised.

[0104] The average detection probability may be determined according to Equation 1, where λ represents the average detection probability. Such average detection probability does not take into account the effect of the detection probability on the number of leak detections or the effect of running intensity on the number of leak detections. λ = (road segment where leak is detected) / (total road segments) (1)

[0105] As an illustrative example, if a model utilizes a 20 percent average detection probability and a particular leak has a 5 percent actual detection probability, the model will likely detect the leak once every five times, but the actual detection probability will cause the system to detect the leak once every 20 times that location is sampled. Thus, because the system is likely to detect the leak once every five samplings but actually detects the leak once every 20 samplings, the system may prematurely rule out the leak (e.g., set the leak to an inactive or non-leaking state). Furthermore, the system may reveal multiple leaks for a particular leak during a period in which the leak persisted but was prematurely ruled out due to non-detection according to the predicted rate.

[0106] In some embodiments, the system implements different detection probabilities for different leaks. Rather than the model assuming one detection probability, various embodiments configure the model to allow the detection probability to be adapted / adjusted. By determining detection probabilities for a subset of leaks (e.g., different detection probabilities for each leak), the system can overcome the problem of leaks that do not have the mean / median detection probability behavior being shut off too early or too late (e.g., set to an inactive or non-leaking state). In some embodiments, the model customizes (e.g., determines) the leak probability for each leak (e.g., a leak for which the model configures a leak state). The model may be configured to allow the leak detection probability for a particular leak to vary between various detection probabilities. The model selects the detection probability that best matches the observed data (e.g., sensor data). Thus, the model selects / determines the leak detection probability rather than using a preselected detection probability across all leaks. Detection probability is a useful way to characterize leaks. For example, larger leaks generally have higher detection probabilities.

[0107] In connection with determining the leak state for a particular leak, the model determines whether the leak is likely still on (e.g., leak-on state) or off (e.g., no-leak state) despite the sensor data not including recent sampling at a particular location (e.g., a subregion associated with a cluster of leaks). For example, the system utilizes an estimate of the leak's historical detection rate. In some embodiments, the model determines the leak state for a leak using long-term historical sensor data and a set of recent sampling data.

[0108] In some embodiments, the model is a Hidden Markov Model (HMM). An HMM assumes that historically, the system can estimate the probability of detection based on past detections (e.g., all past detections) and uses the time period since the last detection to determine the probability of getting the same detection reading consecutively. As an example, an HMM returns a value of 1 as an indication that the leak is active and a value of 0 as an indication that the leak is inactive.

[0109] If the model determines that the leak is almost certainly shut off (e.g., repaired), the system sets the leak status to inactive. If a subsequent leak is detected in the same location as the leak that was set to inactive, the system considers the new leak detection to be associated with a new leak. In some embodiments, the system associates a series of leaks that occurred in the same location / area. The sensitivity of the model in turning off a leak (e.g., considering the leak inactive) may be adjusted between a relatively conservative model where the model may wait too long to turn off a leak (e.g., consider the leak inactive). For example, even if the system knows that the leak is likely to be shut off, a conservative model may be configured to wait for more evidence (e.g., non-detection) before turning off the leak. Such a relatively conservative model therefore creates a delay in the ability to evaluate the leak. Conversely, the model may be configured to turn off leaks more quickly (e.g., turn off leaks more aggressively), which may result in a single leak being split into a series of separate leaks.

[0110] In some embodiments, the system analyzes the leak history to assess whether the leak was turned off too early (e.g., deemed inactive). For example, the system looks at observations made after the leak was turned off to assess whether the leak was turned off too early or whether the observations correspond to a series of separate leaks. The system may use the analysis of the leak history in connection with adjusting models or otherwise providing updates to the leak profile (e.g., information used by customers or regulatory agencies to determine the extent of the leak).

[0111] In some embodiments, the system uses three states to classify inputs to a model used to determine the leak state. The three states include a leak-on state, a no-leak state, and an unsampled state (e.g., no measurements have been collected for a particular location at a particular time). The system queries a model (e.g., an HMM) to determine the leak state. Over a period of time that a particular location remains unsampled, the model smooths the observations between the leak-on state and a no-leak state (e.g., an inactive leak). For example, if a leak is detected and the location remains unsampled for a period of time, during which time the leak probability (e.g., the probability that a leak exists) decreases over time, and eventually the model considers the leak to be inactive and the system sets the leak to a no-leak state.

[0112] In some embodiments, the system uses a model to determine how far or near a leak state change is predicted to occur. For example, the system determines the time at which a leak having a leak-on state is considered inactive and set to a leak state. The predicted time of the leak state transition may be based at least in part on the estimated leak probability. The system can use this predicted time of the leak state transition in connection with determining a trip plan or a sampling plan for a particular location. For example, if the predicted time of the leak state is close (e.g., within a predetermined time threshold), the system can weight a particular location (e.g., a road segment) more heavily in the model for selecting locations to sample during the session. Sampling that location at a time closer to the predicted time of the leak state transition allows for a more accurate determination of the time at which the leak will be shut off. For example, the model has greater uncertainty associated with the leak probability for times closer in time to the predicted leak transition state. Conversely, the system may assign a lower weight to a particular location for which the predicted leak transition time is longer than a threshold period.

[0113] The model may be configured to include a hard break (e.g., six months) for transitioning from an active leak to an inactive leak. For example, if the system determines that a leak is active and the location remains unsampled for the break period, the system considers the leak to be inactive.

[0114] In some embodiments, the system uses the number of detected leaks and the sampling intensity to jointly estimate the detection probability and leak probability for a region. The leak probabilities of various locations within a geographic area may be used to rank the regions based on how bad the system predicts the leak will be. The system can use the ranking in connection with determining a trip plan (e.g., a route / plan for sampling different locations during a session) or updating a map of leaks across a geographic region.

[0115] In various embodiments, the system models the road segment level observation process as a mixture of leaky and non-leaky components, including the detection probability. For example, the system determines the detection probability according to Equations 2-6. P(d i )=λP(d i │n i ,l i =1)+(1-λ)P(d i │n i ,l i =0) (2) l i ~Bernoulli(λ) (3) P(d i │n i ,l i =1)~binomial(n i ,θ i ) (4) P(d i │n i ,l i =0)~binomial(n i ,0) (5) θ i ~Beta (α,β) (6)

[0116] The system may estimate λ, α, and β in Equations 2-6 above using Markov Chain Monte Carlo (MCMC) or Variational Inference (VI), which are commonly available in a variety of programming languages ​​(e.g., Phyton). i (observation) is the number of detections on road segment i, and n i (observation) is the number of passes on road segment i, and l i (partial observation) is the presence of a leak on segment i, λ is the probability that any segment has a leak, and θ i is the probability of detecting a leak if one exists. In some embodiments, the model uses α and β to determine θ, and λ and θ to predict the detected condition.

[0117] In some embodiments, the system determines the leak probability (or leak probability distribution) based at least in part on one or more of the following: the probability that a detected leak is a false positive; the probability that a detected leak is genuine (e.g., the detected leak is actually a leak); the probability that a detected non-leak is a false negative (e.g., the non-leak detection should be the leak that was detected); and the probability that a non-leak is genuine (e.g., the detected non-leak is actually a non-leak). The system uses information (e.g., detection probability, etc.) across multiple road segments to account for false positives or false negatives among detections. For example, the system shares information about the detection rate on particular road segments where leaks are detected to attempt to estimate the number of road segments that do not have leaks detected but actually have leaks (e.g., based on a relatively small sample set of available sensor data for a particular location). The system can model road segments with no detections as coming from a similar detection probability distribution as selected road segments where the system detected leaks.

[0118] The model weights the number of detections against the number of samples (e.g., measurements collected) for a particular location and uses a binomial function to obtain a probability distribution. [Table 1]

[0119] In the example shown in Table 1, the system determines that there are no detections on road segment B. However, the fact that the sensor data shows no detections at all for road segment B does not determine that there is no leak, since the detection probability is not 100 percent. Therefore, the system uses the model to predict the detection probability or leak probability for road segment B, assuming that the sensor data has no detections for road segment B. In some embodiments, the system predicts the detection probability or leak probability for road segment B using the detection probability or leak probability for one or more of road segments A and C. The system can use this information to set the leak status for road segment B. In some embodiments, the system weights the contributions of various road segments used to predict the detection probability for a particular road segment. The weights may be determined based on various factors, such as the distance between the road segment and the particular road segment for which a detection probability is being predicted, location characteristics (e.g., geography, topography, wind profile, time of day when the sample was collected, infrastructure carrying the gas (e.g., type of pipe), etc.). In some embodiments, the system selects road segments to utilize in connection with predicting the detection probability for a particular road segment based on the road segments having a minimum degree of similarity in road segment profile / characteristics.

[0120] The system can utilize the model to estimate the incidence of road segments having leaks within a region while adjusting the sampling intensity, e.g., some locations within a geographic area are sampled more frequently than others (e.g., the locations sampled are determined probabilistically for a particular session).

[0121] 11 illustrates a method for providing a model that predicts leak probability at various locations within a given geographic area, according to various embodiments. In some embodiments, process 1100 is performed, at least in part, by system 100 of FIG.

[0122] In step 1105, the system obtains leakage and road segment passage data collected over a geographic region. The leakage data is generated by the systems described above (e.g., clustering and querying of HMM models).

[0123] In step 1110, the system determines a model for predicting leak probability for one or more subregions within the geographic region. The model may predict leak probability based at least in part on the sensor data, the detection probability, and the collection intensity.

[0124] The model is provided in step 1115. In some embodiments, the system provides the model in step 1115 to another invoked system or service (such as a service configured to determine status for active leaks, inactive leaks, etc., or a service configured to generate a map of leaks over a geographic region).

[0125] At step 1120, a determination is made as to whether process 1100 is complete. In some embodiments, process 1100 is determined to be complete in response to a determination that there is no more leak data to provide, no more air quality measurements to be collected, no more leaks to be analyzed, an administrator indicating that process 1100 is to be paused or stopped, etc. In response to a determination that process 1100 is complete, process 1100 ends. In response to a determination that process 1100 is not complete, process 1100 returns to step 1105.

[0126] 12 illustrates the relationship between the detection probability distribution and the leak probability distribution, according to various embodiments. Graph 1200 shows that the detection probability distribution is coupled to the leak probability distribution. Based on the illustrated example relationship, if the model / system does not consider the probability of detection, the system will not accurately measure the probability of leak at a particular location.

[0127] In some embodiments, the system analyzes sensor data to identify contaminant detections being monitored across a geographic region. The system may cluster the detections across the geographic region into different leaks. Furthermore, the system may determine clusters at the same specific location over a statistically significant time period as different leaks occurring at or near the same location.

[0128] In response to determining that the sensor contains a set of contaminant measurements corresponding to a detection, the system clusters the set of contaminant measurements. The system may set a geographically proximate peak of contaminant observations (e.g., a measurement with the greatest contaminant intensity) as the center of the cluster. The system may set a subregion of the geographic area to be associated with a particular leak (e.g., a particular cluster). For example, the system may consider a cluster to be a subregion defined by an area extending radially from the center of the cluster a predetermined distance. The predetermined distance may be configured to adjust the sensitivity of the model to detect different leaks. If the predetermined distance is set too long, the subregion may be defined large enough to include multiple clusters / leaks. Conversely, if the predetermined distance is set too short, the subregion may exclude observations actually associated with the leak corresponding to the subregion. Furthermore, the system may erroneously identify another leak using such observations outside the subregion. In some embodiments, the predetermined distance is between 10 and 20 meters. In some embodiments, the predetermined distance is 15 meters.

[0129] In connection with identifying leaks, the system isolates detections corresponding to particular source types. For example, to identify leaks (e.g., from a gas pipeline), the system determines detections corresponding to biogenic sources and excludes those detections from a model for identifying leaks across a geographic region.

[0130] According to various embodiments, the system identifies leaks and dynamically updates the corresponding leak status based on one or more of current sensor data or probabilistically / statistically using a detection probability distribution, a leak distribution, and historical sensor data for leaks, etc. The system implements a model that predicts the leak status for a leak. For example, the model classifies the leak as either leak-on or inactive (e.g., blocked). The model may be an HMM. In some embodiments, input to the model is classified according to one of three states: leak-on, inactive or non-leak, and unsampled (e.g., no sensor data has been collected for the corresponding location for a particular session or time point).

[0131] The model used to predict the leak state for a leak is configured so that it is more likely for a leak to transition from a leak-on state (e.g., an active state) to a non-leak state (e.g., an inactive state) than for a leak to transition from a non-leak state to a leak state. Thus, over time, the model eventually causes an active leak to settle to an inactive state. The system can adjust the model to transition more quickly from a leak-on state to a non-leak state when there is no new sensor data for a particular location or when there are no detections from the sensor data at a particular location. For example, the system adjusts the number of samples performed at a location where no detections have been recorded before transitioning an active leak to an inactive leak. In some embodiments, the model takes into account repair data, such as records of repairs in a geographic area, which may be obtained from a third-party service (e.g., a local utility).

[0132] Determining the leak start date and leak end date (e.g., the date / time when an active leak transitioned to an inactive leak) may be used to determine the total amount of contaminant generated during the leak's lifetime. For example, a local government or regulatory agency may be interested in knowing the exact leak lifetime to determine the amount of fine to assess against the utility responsible for the leak. Currently, utilities determine the leak lifetime by extrapolating the start of the leak from the date / time of the last measurement in which the system detected no contaminants. Various embodiments use statistically derived leak start and end dates to accurately predict the leak lifetime, thereby shortening the period for which the utility is considered responsible for the leak (e.g., the statistically predicted true leak lifetime is generally shorter than current methods for measuring leak lifetime).

[0133] Related art systems use peak-to-pass ratios in connection with determining whether to transition an active leak to an inactive leak. The peak-to-pass ratio may be calculated as the number of detections made relative to the number of times the location was sampled. Such systems may also utilize the peak-to-pass ratio for all time sensor data (e.g., samplings taken over a predetermined historical period) versus the peak-to-pass ratio for a set of recent samplings (e.g., samplings taken within a threshold period). Related art systems may then derive possible trends for a particular leak.

[0134] Models according to various embodiments are more responsive and well-supported from a data analysis perspective than less sophisticated related art systems. Sensor data is input into the model, which dynamically updates the leak status of leaks within a geographic area. The system may determine the leak status probabilistically. As one example, the system may identify a newly detected leak based on observed sensor data and a corresponding detection probability distribution associated with the location where the leak is detected. As another example, the system may decide to transition an active leak to an inactive leak based on the detection probability, leak probability, and sensor data (e.g., both historical and recent sensor data).

[0135] In some embodiments, if a first leak at a particular location transitions from a leak-on state to an inactive state and a second leak is subsequently detected at the same location, the system considers the first and second leaks to be separate leaks. However, the system may associate the first and second leaks based on their occurrence at the same location, for example, to provide an indication of where infrastructure (e.g., a gas pipeline) is prone to leaks.

[0136] The system generates a map of a geographic area (e.g., an area the system is contracted to monitor over a particular period of time) in which various leaks are indicated. The map may include an indication of whether a particular leak is an active leak or a leak that recently became inactive (e.g., a leak that went inactive within a threshold period of time). The map may represent detections of particular pollutants by road segment. For example, the system may display pollutant detections by road segment in a manner that labels the detections according to pollutant intensity (e.g., low-intensity detections may be color-coded green, medium-intensity detections may be color-coded yellow, and high-intensity detections may be color-coded red, or pollutant intensity may be color-coded along a predetermined spectrum).

[0137] 13 illustrates a method for determining a leakage condition for a cluster according to various embodiments. In some embodiments, process 1300 is performed at least in part by system 100 of FIG.

[0138] In step 1305, the system acquires sensor data. The sensor data may be collected by one or more mobile sensors (e.g., a sensor platform mounted on a vehicle). For example, the sensor data corresponds to a particular sampling, such as a travel date, when a vehicle traveled a corresponding road segment and performed the sampling (e.g., collected air quality / pollutant measurements).

[0139] In step 1310, the system detects peaks in the contaminant signal in the sensor data. The contaminant signal may be the particular contaminant being monitored (e.g., methane, ethane, etc.). Process 1300 may be run multiple times, each time corresponding to the analysis of a different contaminant. In some embodiments, the system detects peaks based on determining that the strength of the contaminant signal exceeds a threshold level (e.g., the sensor noise baseline or a threshold set for the sensor noise baseline (e.g., three times the sensor noise baseline)).

[0140] In step 1315, the system determines a sub-region of a predetermined size centered on the location corresponding to the peak. For example, if the system determines that the peak corresponds to a leak and that the leak is not an existing leak, the system may determine to form a new sub-region for the new cluster of sensor data. The sub-region of a predetermined size may be a circular area radiating from a central location (e.g., the location of the peak), for example, the predetermined sub-region is an area radiating out to 15 meters (or other predetermined distance threshold) from the center.

[0141] In step 1320, the sub-region is established as a cluster for a particular leak emission source. The system considers the sub-region to correspond to a leak and collects sensor data for the sub-region (e.g., when the area is sampled) to monitor the leak, such as to determine whether the leak is persistent or inactive (e.g., repaired).

[0142] In step 1325, the system monitors the sensor data for the cluster. As sampled sensor data is collected within the subregion, the system updates the model and updates or maintains the leak state (e.g., leak state, non-leak state, unsampled state) based on the sensor data.

[0143] In step 1330, the system determines a leak state based on the monitored sensor data. For example, the system determines whether to maintain the leak state (e.g., by querying a model for a predicted leak classification or leak probability) or whether to set the state to a non-leak or unsampled state. The system may decide to set the state to an unsampled state in response to determining that no sampling has been performed within the subregion (or the degree of sampling is below a sampling threshold).

[0144] In step 1335, the system determines whether to update the leak status. For example, the system determines whether to continue monitoring for a leak. In response to determining a leak status, process 1300 may return to step 1325, and the process may repeat steps 1325-1330 until the system determines that the leak status is no longer to be updated.

[0145] In step 1340, the system provides the leak status for the cluster. The system can provide the leak status to another system or service that invoked process 1300 (e.g., a mapping service that generates a map of leaks over the geographic area being monitored).

[0146] At step 1345, a determination is made as to whether process 1300 is complete. In some embodiments, process 1300 is determined to be complete in response to a determination that there is no more leak data to provide, no more air quality measurements to be collected, no more leaks to be analyzed or classified, no more peaks detected, an administrator indicating that process 1300 is to be paused or stopped, etc. In response to a determination that process 1300 is complete, process 1300 ends. In response to a determination that process 1300 is not complete, process 1300 returns to step 1305.

[0147] 14 illustrates a method for determining the leak status of a particular leak, according to various embodiments. In some embodiments, process 1400 is performed, at least in part, by system 100 of FIG.

[0148] In step 1405, the system acquires sensor data. The sensor data may be collected by one or more mobile sensors (e.g., a sensor platform mounted on a vehicle). For example, the sensor data corresponds to a particular sampling, such as a travel date, when a vehicle traveled a corresponding road segment and performed the sampling (e.g., collected air quality / pollutant measurements).

[0149] In step 1410, the system selects locations within the geographic region being monitored. For example, the system determines locations that correspond to sensor data (e.g., locations where a mobile sensor platform sampled / collected air quality measurements).

[0150] In step 1415, the system sets the state of the detection (e.g., sensor data) at the selected location to one of a leak state, a non-leak state, or an unsampled state. The system sets the state based at least in part on the sensor data. For example, the system may consider the state to be an unsampled state in response to determining that the location has not been sampled for a predetermined time threshold (or sampling at the location is below a predetermined sampling threshold). As another example, the system may decide to set the state to a leak state in response to determining that a contaminant signal (e.g., a contaminant signal having an intensity greater than an intensity threshold (e.g., a threshold set relative to a sensor noise baseline)) has been detected. As another example, the system may decide to set the state to a non-leak state in response to determining that no contaminant signal was detected or that the detected contaminant signal was below an intensity threshold (e.g., a level at which the contaminant signal can be attributed to inherent sensor noise).

[0151] In step 1420, the system queries the model for the estimated leak condition. For example, the system queries the model for a predicted leak classification or a predicted leak probability (e.g., the probability that a leak exists at a selected location). The model may be a hidden Markov model (HMM) that predicts the leak condition using historical and current sensor data for the selected location and / or locations proximate to the selected location. Locations proximate to the selected location may correspond to locations within a threshold distance of the selected location and / or locations deemed to have similar location characteristics (e.g., terrain, wind, underlying infrastructure into which a leak may be released (e.g., type of pipe carrying gas), etc.).

[0152] At step 1425, the system determines whether to estimate a leak condition at additional locations within the geographic area. For example, the system may iterate over various locations (e.g., road segments, census tracts, etc.) in connection with mapping leaks across a geographic region. In response to determining that a leak condition is estimated at additional locations, process 1400 returns to step 1410, and process 1400 repeats steps 1410-1420 until the system does not estimate a leak condition at any additional locations. Conversely, in response to determining that the system does not estimate a leak condition at any additional locations, process 1400 proceeds to step 1430.

[0153] At step 1430, the system provides the leak status. In some embodiments, the system provides the leak status to another system or service that invoked process 1400 (e.g., a mapping service that maps leaks over the geographic area being monitored). For example, the system provides an alert to a user or other system or service that a new leak has been detected. As another example, the system updates a map of leaks detected over the geographic area being monitored. The system may use the leak status indication in connection with determining a trip plan (e.g., a plan for sampling at specific locations) or calling for repairs.

[0154] At step 1430, a determination is made as to whether process 1400 is complete. In some embodiments, process 1400 is determined to be complete in response to a determination that there is no more leak data to provide, no more air quality measurements to be collected, no more leaks to be analyzed, an administrator indicating that process 1400 is to be paused or stopped, etc. In response to a determination that process 1400 is complete, process 1400 ends. In response to a determination that process 1400 is not complete, process 1400 returns to step 1405.

[0155] 15 illustrates a method for determining the leak status of a particular leak, according to various embodiments. In some embodiments, process 1500 is performed, at least in part, by system 100 of FIG.

[0156] In step 1505, the system obtains an indication to determine whether an active leak is predicted to be transitioned to a non-leak state. The leak probability for a particular leak may decrease over time if there are no further detections in the corresponding location (e.g., a sub-region for a leak cluster).

[0157] In step 1510, the system acquires leak data for active leaks, for example, the system acquires leak measurements (e.g., leak detected or no leak detected) over a predetermined period of time.

[0158] In step 1515, the system determines the leak location of the active leak, for example, the system determines a sub-region associated with the active leak or the probable source (e.g., emission source) of the leak.

[0159] In step 1520, the system determines whether the sensor data for the leak location includes recent sensor data. The system may determine whether the leak location was sampled within a predetermined period of time. The predetermined period may be configurable, such as to adjust the sensitivity of the model in the absence of leak detection (e.g., the sensitivity of the model as to whether it classifies a leak as inactive).

[0160] In response to determining in step 1520 that the leak location (e.g., a subregion corresponding to a cluster of leaks) has not been sampled recently (e.g., there is no recent sensor data for the leak location), process 1500 proceeds to step 1525, where the system obtains a leak probability distribution for the leak. The leak probability distribution may be determined based at least in part on historical input states (e.g., leak state, non-leak state, unsampled state) for the data at the leak location.

[0161] In response to determining in step 1520 that the leak location has been recently sampled (e.g., there is recent sensor data for the leak location), process 1500 proceeds to step 1530, where the system obtains sensor data for the leak location. The system may obtain historical sensor data, including recent sensor data collected for the leak location.

[0162] In step 1535, the system determines whether the active leak is predicted to have transitioned to a non-leak state based at least in part on the leak probability distribution. For example, the system queries a model (e.g., a hidden Markov model) for a probable leak state (e.g., a predicted leak classification). For example, the model determines whether the leak is likely still on despite not being detected recently, or whether the leak is likely to have been deactivated (e.g., repaired and set as an inactive leak). The system may determine a predicted leak classification even if the leak has not been seen recently, based on, for example, an estimate of historical detection rates for the leak or similar leaks (e.g., leaks occurring in similar locations (e.g., similar terrain, weather, wind profiles, underlying infrastructure carrying contaminants (e.g., pipe type (cast iron, plastic, etc.))).

[0163] In response to determining that the active leak is predicted to have transitioned to a non-leak state (e.g., the predicted leak probability is less than a predetermined leak threshold), process 1500 proceeds to step 1545. Conversely, in response to determining that the active leak is not predicted to have transitioned to a non-leak state (e.g., become inactive), process 1500 proceeds to step 1550.

[0164] At step 1540, the system determines whether the sensor data includes a leak signal. For example, the system determines whether the sensor data for the leak location indicates the presence of a particular contaminant (e.g., a monitored contaminant, such as methane). In some embodiments, the system determines that the sensor data includes a leak signal in response to determining that the detected contaminant signal exceeds a threshold (e.g., a predetermined threshold or a predetermined degree relative to the sensor noise baseline (e.g., three times the sensor noise baseline)). In response to determining that the sensor data does not include a leak signal, process 1500 proceeds to step 1545. Conversely, in response to determining that the sensor data does include a leak signal, process 1500 proceeds to step 1555.

[0165] In step 1545, the system determines that the active leak state is set to a no leak state.

[0166] In step 1555, the system determines that the active leak condition remains a leak condition.

[0167] At step 1560, the system provides an indication of the leak status. The indication of the leak status may be provided to another system or service, such as the system / service that invoked process 1500 (e.g., a mapping service that maps leaks across the geographic region being monitored) or a third party system (e.g., a regulatory agency or customer). In some embodiments, the system provides the indication of the leak status to a regulatory agency or customer in connection with monitoring leaks across the geographic area.

[0168] At step 1565, a determination is made as to whether process 1500 is complete. In some embodiments, process 1500 is determined to be complete in response to a determination that there is no more leak data to be provided, no more air quality measurements to be collected, no more leaks to be analyzed, an administrator indicating that process 1500 is to be paused or stopped, etc. In response to a determination that process 1500 is complete, process 1500 ends. In response to a determination that process 1500 is not complete, process 1500 returns to step 1505.

[0169] 16 illustrates a method for associating detected peaks in sensor data with new or existing clusters associated with particular leaks, according to various embodiments. In some embodiments, process 1600 is performed at least in part by system 100 of FIG.

[0170] In step 1605, the system acquires sensor data. The sensor data may be collected by one or more mobile sensors (e.g., a sensor platform mounted on a vehicle). For example, the sensor data corresponds to a particular sampling, such as a travel date, when a vehicle traveled a corresponding road segment and performed the sampling (e.g., collected air quality / pollutant measurements).

[0171] In step 1610, the system detects peaks in the contaminant signal in the sensor data. The contaminant signal may be a particular contaminant being monitored (e.g., methane, ethane, etc.), and process 1600 may be run multiple times, each time corresponding to the analysis of a different contaminant. In some embodiments, the system detects peaks based on a determination that the strength of the contaminant signal exceeds a threshold level (e.g., the sensor noise baseline or a threshold set for the sensor noise baseline (e.g., three times the sensor noise baseline)).

[0172] In step 1615, the system determines whether the geographic location for the peak in the contaminant signal is part of an existing cluster. For example, the system determines whether the geographic location corresponding to the sampling of sensor data where the contaminant signal peak was detected is within a sub-region associated with the cluster. The sub-region may be of a predetermined size, such as an area extending 15 meters radially from the center of the cluster. Various other cluster sizes may also be implemented. For example, the size of the sub-region for a particular cluster may be adjustable to control the sensitivity of the detection model. Making the sub-region too large may cause multiple leaks to be clustered and identified as a single leak. Conversely, making the sub-region too small may cause contaminant signal leaks sampled near the outside of the sub-region to be classified as separate leaks (e.g., false positives).

[0173] In response to determining that the location for the peak in the contaminant signal is not part of an existing cluster, process 1600 proceeds to step 1620, where the system generates a leak suggestion. The system may associate a start date with the leak suggestion, which corresponds to the date in the sensor data on which the peak was detected / identified (e.g., the first day the contaminant signal was detected for a particular location for the apparent leak). Process 1600 then proceeds to step 1625, where the leak suggestion is provided. For example, the system may provide an alert to a user or other system or service that a new leak has been detected. As another example, the system updates a map of detected leaks across the geographic area being monitored. The system may use the new leak suggestion in connection with determining a driving plan (e.g., a plan for sampling at specific locations) or calling for repairs.

[0174] In response to determining that the location for the peak in the pollutant signal is part of an existing cluster, process 1600 proceeds to step 1630, where the system associates the pollutant signal leak (e.g., sensor data for a particular location) with the existing cluster. The system determines a cluster having a sub-region that includes the location for the sensor data in which the pollutant signal peak was detected. Then, in step 1635, the system provides information about the existing cluster. For example, the system provides an alert to a user or other system or service that the corresponding leak persists. As another example, the system updates a map of detected leaks across the geographic area being monitored. The system may use the leak indication in connection with determining a driving plan (e.g., a plan for sampling at a particular location) or calling for repairs.

[0175] At step 1640, a determination is made as to whether process 1600 is complete. In some embodiments, process 1600 is determined to be complete in response to a determination that there is no more leak data to provide, no more air quality measurements to be collected, no more leaks to be analyzed, an administrator indicating that process 1600 is to be paused or stopped, etc. In response to a determination that process 1600 is complete, process 1600 ends. In response to a determination that process 1600 is not complete, process 1600 returns to step 1605.

[0176] 17 illustrates a method for detecting a leak, according to various embodiments. In some embodiments, a process 1700 is performed, at least in part, by the system 100 of FIG.

[0177] In some embodiments, process 1700 is performed for one or more contaminants being monitored. For example, process 1700 is performed to control the activation or deactivation of a leak indicator for methane. Various other contaminants may also be monitored.

[0178] In step 1705, the system receives a first information stream indicative of a leak condition. In some embodiments, the sensor data comprising the first information stream is collected by one or more mobile sensors (e.g., collected by a sensor platform attached to a vehicle traveling on a predetermined driving plan). The first information stream may correspond to information for a particular location (e.g., a sub-region corresponding to a cluster of peaks in the detected leak signal).

[0179] In step 1710, the system determines that an initial leak condition exists. The system determines that an initial leak condition exists at a particular location (e.g., a sub-region associated with a cluster of leak detections corresponding to a single emission / leak source) based at least in part on the first information stream. The system may determine that an initial leak condition exists based at least in part on a determination that the location at which the leak detections were collected does not correspond to a cluster associated with an existing leak.

[0180] In some embodiments, the system may determine that an initial leak condition exists based at least in part on querying a model for a prediction of a leak classification, such as a leak probability. If the leak probability exceeds a predetermined leak threshold, the system considers a leak to exist in the corresponding location / sub-region within the geographic region being monitored.

[0181] In step 1715, the system receives a second information stream. The second information stream indicates a no-leak condition (e.g., for a particular location (e.g., a sub-region relative to a cluster) associated with the initial leak). For example, the sensor data collected at the particular location / sub-region associated with the initial leak does not contain a contaminant signal (e.g., there is no peak for a particular contaminant). The system may consider the sensor data to not contain a contaminant signal if the detected contaminant has an intensity less than a predetermined threshold (e.g., a sensor noise baseline).

[0182] At step 1720, the system determines that the leak condition has ended using a statistical model based at least in part on the first information stream and the second information stream. For example, the model provides a predicted leak classification or a predicted leak probability. In response to querying the model to obtain the predicted leak classification or the predicted leak probability, the system determines whether the leak condition has ended.

[0183] In some embodiments, the model is a Hidden Markov Model (HMM). The HMM receives inputs corresponding to sampling of a particular location or subregion (e.g., a subregion associated with a leak, such as an area extending 15 meters radially from the center of a cluster). Inputs to the HMM are classified according to one of three states: leak, non-leak (or no leak), and unsampled (e.g., the location has not been visited). In response to querying the leak classification (e.g., based on a query from another service or according to periodic updates of leak indications or a map of leaks over a geographic area), the HMM provides a predicted leak classification as either a leak state or a non-leak state.

[0184] In step 1725, the system receives a third information stream. The third information stream indicates a leak condition (e.g., for a particular location (e.g., a subregion relative to a cluster) associated with the initial leak). The system may determine that the third information stream indicates a leak condition based on the presence of a leak signal (e.g., a signal for a particular contaminant being monitored, such as methane) having an intensity that exceeds the sensor noise baseline by a predetermined threshold (e.g., the sensor noise baseline or a threshold set relative to the baseline (e.g., three times the sensor noise baseline), etc.).

[0185] In step 1730, the system determines that a new leak exists. The new leak is considered different from the initial leak. The system determines that a new leak exists based at least in part on the third information stream. In some embodiments, the system determines that a new leak exists because the initial leak has been closed or otherwise inactive.

[0186] In some embodiments, the system queries a model (e.g., an HMM) to determine whether the leak indicated by the third information stream is related to the leak indicated by the first information stream (e.g., whether the system prematurely deactivated and set the first leak to inactive).

[0187] In some embodiments, the system determines that the leak indicated by the third information stream is different from the leak indicated by the first stream based at least in part on determining that repairs were performed to fix the initial leak. For example, the system obtains repair data, such as data received from a third party (e.g., data provided by a utility that manages the infrastructure through which the leak was released).

[0188] In response to determining that a new leak exists, the system provides an indication of the new leak. For example, the system provides an alert to a user or other system or service that a new leak has been detected. As another example, the system updates a map of detected leaks over the geographic area being monitored. The system may use the indication of the new leak in connection with determining a trip plan (e.g., a plan for sampling at specific locations) or calling for repairs.

[0189] At step 1735, a determination is made as to whether process 1700 is complete. In some embodiments, process 1700 is determined to be complete in response to a determination that there is no more leak data to provide, no more air quality measurements to be collected, no more leaks to be analyzed, an administrator indicating that process 1700 is to be paused or stopped, etc. In response to a determination that process 1700 is complete, process 1700 ends. In response to a determination that process 1700 is not complete, process 1700 returns to step 1705.

[0190] 18 illustrates a method for providing leakage data according to various embodiments. In some embodiments, process 1700 is performed at least in part by system 100 of FIG.

[0191] In step 1805, the system obtains an instruction to provide leak data for a particular leak.

[0192] In step 1810, the system determines the start date of the leak. The system determines the start date of the leak based on the earliest date that a peak in the contaminant signal was observed in the sensor data. For example, the system determines the date that the system identified the leak, such as based on identifying a cluster.

[0193] In step 1815, the system determines an end date for the leak. In some embodiments, the system determines the end date for the leak based on the earliest date on which the system probabilistically determines that the leak has ended. The system may probabilistically determine that the leak has ended based on sensor data including a set of one or more non-leak condition detections. Additionally or alternatively, the system may probabilistically determine that the leak has ended based on an area associated with the leak remaining unsampled for a period of time. For example, the system determines a date on which the leak is predicted to be resolved (e.g., repaired by a utility) based on the leak probability and sensor data (which may include a set of leak detections, a set of non-leak detections, or a set of values ​​associated with the area being unsampled).

[0194] In some embodiments, the system determines to transition the leak to an inactive state (e.g., a non-leak state) in response to determining that the leak no longer exists. As an example, the system determines that the leak no longer exists based at least in part on sensor data and / or third-party service data (e.g., repair data indicative of repair activity). The system may probabilistically determine that the leak no longer exists based at least in part on a model of leak probability. The model of leak probability predicts the likelihood of a leak being present based at least in part on sensor data (e.g., a set of leak detections, a set of non-leak state detections) or lack thereof (e.g., an area proximate to the leak not being sampled for a period of time). The system may query the model for a prediction of whether a leak is present. In some embodiments, the system updates or maintains the leak state based at least in part on the prediction.

[0195] In step 1820, the system provides the leak data. The leak data may be provided to another system or service, such as the system / service that invoked process 1800 (e.g., a mapping service that maps leaks across the geographic region being monitored) or a third-party system (e.g., a regulatory agency or customer). In some embodiments, the system provides the leak data to a regulatory agency or customer in connection with monitoring leaks across the geographic area. The system may calculate the amount of carbon (or other pollutants) emitted to the atmosphere based at least in part on the predicted start date and the predicted end date. Regulatory agencies and customers may use the amount of carbon in connection with calculating fines or other compensation owed to the customer (or utility).

[0196] At step 1825, a determination is made as to whether process 1800 is complete. In some embodiments, process 1800 is determined to be complete in response to a determination that there is no more leak data to provide, no more air quality measurements to be collected, an administrator indicating that process 1800 is paused or stopped, etc. In response to a determination that process 1800 is complete, process 1800 ends. In response to a determination that process 1800 is not complete, process 1800 returns to step 1805.

[0197] Various example embodiments described herein are described with reference to flowcharts. While the examples may include some steps performed in a particular order, according to various embodiments, various steps may be performed in different orders and / or various steps may be combined into a single step or performed in parallel.

[0198] Although the above-described embodiments have been described in some detail for ease of understanding, the invention is not limited to the details provided. There are many alternative ways of implementing the invention. The disclosed embodiments are illustrative and are not intended to be limiting.

Claims

1. 1. A system for detecting leaks near a source of an emission, comprising:

1. A processor, comprising: receiving a first information stream indicative of a leak condition from one or more movement sensors; determining that an initial leak condition exists based at least in part on the first information stream indicative of the leak condition; receiving a second information stream indicative of a non-leakage condition; determining that the leak condition has terminated using a statistical model based at least in part on the first information stream and the second information stream; receiving a third information stream indicative of the leak condition; a processor configured to determine that a new leak condition exists, the new leak condition being different from the initial leak condition; a memory coupled to the processor and configured to provide instructions to the processor; A system comprising:

2. 10. The system of claim 1, further comprising a communications interface configured to receive sensor data from one or more motion sensors; The system, wherein the first information stream, the second information stream, and the third information stream are obtained based at least in part on the sensor data.

3. 10. The system of claim 1, wherein the initial leak condition exists at a specific location within a geographic area monitored by the one or more mobile sensors.

4. 4. The system of claim 3, wherein the processor further comprises: The system is configured to determine that the state at the particular location corresponds to an unsampled state at a particular time.

5. 5. The system of claim 4, wherein the statistical model is used to detect that the leak condition has ended based at least in part on time series data including a first portion of data corresponding to the leak condition, a second portion of data corresponding to the non-leak condition, and a third portion of data corresponding to the unsampled condition.

6. The system of claim 3 , wherein the particular location corresponds to a particular region corresponding to a cluster of sensor data indicative.

7. 10. The system of claim 1, the initial leak condition exists at a particular location within the geographic region monitored by the one or more mobile sensors; The system wherein the state of the particular location is probabilistically determined to switch from the leak state to a non-leak state based at least in part on the particular location not being sampled for a period of time.

8. 10. The system of claim 1, wherein the first information stream indicative of the leak condition is obtained based at least in part on performing clustering on sensor data collected in a geographic region over a period of time.

9. 9. The system of claim 8, wherein performing the clustering on the sensor data includes determining clusters of measurements collected by the sensor data around the emission source.

10. 10. The system of claim 1, wherein the processor is further configured to implement a Hidden Markov Model (HMM) to determine whether a condition at a particular location corresponds to the leak condition or the non-leak condition.

11. 10. The system of claim 1, wherein the processor is further configured to provide an indication of when a particular leak begins and when a particular leak ends.

12. 10. The system of claim 1, wherein the processor is further configured to classify a source type associated with the leak into a biogenic source type or a pyrolytic source type.

13. 13. The system of claim 12, wherein the processor classifies the source type as the biogenic source type or the pyrolytic source type based at least in part on a determination of whether sensor data near the emission source indicates the presence of ethane.

14. The system of claim 1 , wherein the initial leak condition is determined to exist based at least in part on a probability of detection.

15. 15. The system of claim 14, wherein the probability of detection includes a leak component corresponding to the probability of detecting a leak and a non-leak component corresponding to the probability of detecting a non-leak.

16. 10. The system of claim 1, wherein the processor further comprises: determining whether a condition at a particular location is predicted to transition from said leak condition to said non-leak condition at a particular time; The system is configured to update a sampling plan to cause the one or more mobile sensors to sample the particular location within a predetermined time period from the particular time.

17. 10. The system of claim 1, wherein the processor further comprises: obtaining repair data from a third-party service indicative of repair activity within geographic locations from which the one or more mobile sensors collect sensor data; determining whether the repair data indicates that a repair was performed within a vicinity of the emission source between when the first information stream was received and when the third information stream was received; In response to determining that the repair has been performed within a vicinity of the emission source, the system is configured to determine that the leak indicated by the third information stream is different from the initial leak.

18. 10. The system of claim 1, wherein the processor further comprises: obtaining repair data from a third-party service indicative of repair activity within geographic locations from which the one or more mobile sensors collect sensor data; determining whether the repair data indicates that a repair was performed within a vicinity of the emission source between when the first information stream was received and when the third information stream was received; In response to determining that the repair has not been performed within a vicinity of the emission source, the system is configured to determine that the leak indicated by the third information stream is not different from the initial leak.

19. 1. A method for detecting a leak near an emission source, comprising: receiving a first information stream indicative of a leak condition from one or more movement sensors; determining that an initial leak condition exists based at least in part on the first information stream indicative of the leak condition; receiving a second information stream indicative of a non-leakage condition; determining that the leak condition has terminated using a statistical model based at least in part on the first information stream and the second information stream; receiving a third information stream indicative of the leak condition; determining that a new leak condition exists, the new leak condition being different from the initial leak condition; A method comprising:

20. 1. A computer program product for detecting leaks near an emission source, the computer program product embodied in a tangible computer-readable storage medium; computer instructions for receiving a first information stream indicative of a leak condition from one or more movement sensors; computer instructions for determining that an initial leak condition exists based at least in part on the first information stream indicative of the leak condition; computer instructions for receiving a second information stream indicative of a no-leak condition; computer instructions for determining, based at least in part on the first information stream and the second information stream, using a statistical model, that the leak condition has ended; computer instructions for receiving a third information stream indicative of the leak condition; computer instructions for determining that a new leak condition exists, the new leak condition being different from the initial leak condition; A computer program product comprising:

21. 1. A method for classifying a gas signal, comprising: receiving sensor data collected over a geographic region from one or more mobile sensors; Detecting a first gas signal in the sensor data; determining a source type of the first gas based at least in part on determining whether the sensor data includes a signal of another contaminant; providing said source type; A method comprising:

22. 22. The method of claim 21, wherein the first gas signal corresponds to a methane gas signal.

23. 23. The system of claim 22, wherein the other contaminant corresponds to ethane.

24. 24. The method of claim 23, wherein the source type is deemed to correspond to a biogenic source type in response to determining that the sensor data includes the methane gas signal without the presence of an ethane gas signal.

25. 24. The method of claim 23, wherein the source type is deemed to correspond to a pyrolytic origin source type in response to determining that the sensor data includes the methane gas signal and the ethane gas signal.

26. 24. The method of claim 23, wherein the sensor data is deemed to include an ethane signal in response to determining that an ethane measurement in the sensor data exceeds a noise baseline by a predetermined amount.

27. 27. The method of claim 26, wherein the predetermined degree corresponds to at least 300% of the noise baseline.

28. 28. The method of claim 27, wherein the noise baseline is a moving baseline over a predetermined time window.

29. 1. A method for determining a ranking of sub-regions within a geographic region based on incidence of gas leaks, comprising: obtaining sensor data collected over a geographic region, the sensor data being collected by one or more mobile sensors; determining a model for predicting leakage probability for one or more sub-regions within the geographic region based at least in part on the number of clusters, the number of detections per cluster, and the collection intensity; providing said model; A method comprising:

30. 30. The system of claim 29, wherein the gas corresponds to methane.

31. 30. The system of claim 29, wherein the one or more sub-regions include one or more road segments.

32. 30. The method of claim 29, wherein the probability of detection includes a leak component corresponding to the probability of detecting a leak and a non-leak component corresponding to the probability of detecting a non-leak.

33. 33. The method of claim 32, wherein the probability of detection comprises a probability of detecting the leak when the leak is not present.

Images