Leak detection event aggregation and ranking systems and methods
In some embodiments, data from multiple vehicle-based natural gas leak detection survey runs are used by computer-implemented machine learning systems to generate a list of natural gas leaks ranked by hazard level. A risk model embodies training data having known hazard levels, and is used to classify newly-discovered leaks. Hazard levels may be expressed by continuous variables, and/or probabilities that a given leak fits within a predefined category of hazard (e.g. Grades 1-3). Each leak is represented by a cluster of leak indications (peaks) originating from a common leak source. Hazard-predictive features may include maximum, minimum, mean, and/or median CH4/amplitude of aggregated leak indications; estimated leak flow rate, determined from an average of leak indications in a cluster; likelihood of leak being natural gas based on other indicator data (e.g. ethane concentration); probability the leak was detected on a given pass; and estimated distance to leak source.
1. A system comprising at least one hardware processor and a memory storing instructions which, when executed, cause the at least one hardware processor to:
receive collected data including gas concentration and location values collected by a vehicle-borne gas concentration measurement device configured to perform a sequence of geospatially-referenced mobile gas concentration measurements along one or more survey paths;
automatically analyze the collected data to identify a plurality of natural gas leaks; and
generate a list of leaks sorted by hazard level by applying a risk model to the collected data.
2. The system of claim 1 , wherein analyzing the collected data comprises applying a filter to a set of leak detections to filter out a subset of leak detections, the filter including at least one filter selected from a filter based on a background level of natural gas, and a filter based on an inter-peak distance for leak detections.
3. The system of claim 1 , wherein analyzing the collected data comprises spatially aggregating leak indications of the collected data into a plurality of clusters, each cluster comprising a plurality of leak indications, each cluster representing a corresponding leak.
4. The system of claim 3 , wherein generating the plurality of clusters comprises performing density-based spatial clustering using a maximum distance scale and a minimum number of points per cluster as input parameters to distinguish between members and non-members of each cluster.
5. The system of claim 3 , wherein generating the plurality of clusters comprises carrying out Markov clustering by performing a random walk through points representing leak indications to distinguish between members and non-members of each cluster.
6. The system of claim 3 , wherein generating the plurality of clusters comprises employing wind directions values corresponding to leak indications to distinguish between members and non-members of each cluster.
7. The system of claim 1 , wherein the risk model embodies training data characterizing known leaks and associated hazard levels.
8. The system of claim 1 , wherein applying the risk model comprises determining values for a plurality of hazard-predictive features for a leak according to the collected data, and determining a hazard level for the leak according to the determined values.
9. The system of claim 8 , wherein the plurality of hazard-predictive features includes a maximum concentration, a minimum concentration, and a representative concentration for the leak, determined from a cluster of leak indications assigned to the leak.
10. The system of claim 8 , wherein the plurality of hazard-predictive features includes an estimated leak flow rate for the leak.
11. The system of claim 8 , wherein the plurality of hazard-predictive features includes an estimated likelihood of the leak being a natural gas leak, the estimated likelihood determined according to a relationship between ethane and methane concentration values in the collected data.
12. The system of claim 8 , wherein the plurality of hazard-predictive features includes an estimated distance between a leak indication and a corresponding leak source.
13. The system of claim 8 , wherein the risk model is selected from a linear regression model yielding continuous values representing hazard levels, a logistic regression model yielding probabilities that leaks fit within predefined categories of hazard, and a random forest model.
14. The system of claim 1 , wherein the hazard level for a leak represents an estimated probability that the leak fits within a predefined category of hazard.
15. A method comprising employing a system comprising at least one hardware processor and a memory storing instructions which, when executed, cause the at least one hardware processor to:
receive collected data including gas concentration and location values collected by a vehicle-borne gas concentration measurement device configured to perform a sequence of geospatially-referenced mobile gas concentration measurements along one or more survey paths;
automatically analyze the collected data to identify a plurality of natural gas leaks; and
generate a list of leaks sorted by hazard level by applying a risk model to the collected data.
16. A non-transitory computer-readable medium encoding instructions which, when executed by at least one hardware processor of a computer system, cause the computer system to:
receive collected data including gas concentration and location values collected by a vehicle-borne gas concentration measurement device configured to perform a sequence of geospatially-referenced mobile gas concentration measurements along one or more survey paths;
automatically analyze the collected data to identify a plurality of natural gas leaks; and
generate a list of leaks sorted by hazard level by applying a risk model to the collected data.