Event Map Decay Function for Critical Event Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for detecting critical events in public places inadequately account for the time profile of sensor measurement values, leading to increased detection of measurement artifacts and difficulties in localizing moving objects, and face challenges in data representation and processing with multi-sensor applications.
Innovation Solution
A method involving the creation of event maps with nodes arranged in a grid, where sensor readings are mapped to nodes based on their measurement areas, and a decay function is applied to intensity values over time to determine event values, with threshold values specific to each node, allowing for weighted accumulation and detection of critical events.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If sensor measurement values are evaluated without considering time profile, then detection speed is improved, but measurement artifacts increase and detection precision deteriorates
Solution Approach 1:
The patent applies dynamics by introducing time-dependent decay functions that dynamically adjust the weight of historical sensor data. The decay function λ(t) = e^(-t/τ) dynamically reduces the influence of older measurements while maintaining sensitivity to recent events, enabling the system to adapt to changing conditions without sacrificing detection precision
Solution Approach 2:
The patent introduces an intermediary accumulation value A(x, t) that mediates between raw sensor measurements and final event detection. This accumulation value integrates temporal information through weighted summation of decayed intensity values, serving as an intermediate representation that preserves time profile information while enabling efficient detection
2Reliability
If multiple sensors are used to detect critical events, then detection reliability is improved, but data representation and processing complexity increases
Solution Approach 1:
The patent merges data from multiple sensors by mapping their measurement values to a common spatial grid structure. Event maps from different sensors are overlaid and accumulated at corresponding grid positions, combining information from multiple sources while maintaining a unified data representation that simplifies processing
Solution Approach 2:
The patent transforms sensor data from raw measurement values into standardized event intensity values that are normalized and scaled. This parameter transformation allows data from different sensor types and scales to be meaningfully combined and compared, reducing processing complexity while maintaining detection reliability
3Productivity
If event accumulation is performed without decay function, then accumulation speed is improved, but detection precision deteriorates due to outdated events
Solution Approach 1:
The patent implements periodic action through the decay function that continuously reduces the weight of accumulated events over time. This creates a natural expiration mechanism where old events gradually lose influence, allowing the accumulation process to remain computationally efficient while automatically prioritizing recent and relevant events
Solution Approach 2:
The decay function dynamically changes the parameter of event weight over time, transforming static accumulation into a time-aware process. The exponential decay parameter λ(t) = e^(-t/τ) automatically adjusts the contribution of each event based on its age, maintaining precision without sacrificing accumulation speed
Data Source
Figure 1
Figure 2
Figure 3~4b
AI summary
Method for detecting safety-critical events (EC) occurring at a specific detection time (td) in areas to be monitored in public places, a) wherein sensor measurements (m1, m2) are determined using individual sensors (S1, S2), and several event maps (C1, C2) are created, each with a number of nodes (Nx, Ny), wherein each of the nodes (Nx, Ny) or a plurality of the nodes (Nx, Ny) is assigned to an area in the measurement range of at least one of the sensors (S1, S2) on at least one event map (C1, C2), b) wherein a common map (C) with a number of nodes (Mz) is specified, wherein each of the nodes (Mz) or a plurality of the nodes of the common map (C) is assigned to an area in the measurement range of at least one of the sensors (S1, S2), c) wherein a mapping (T1, T2) onto the common map (C) is specified in advance for each of the individual event maps (C1, C2), which in particular displays measured values (m1,m2) relating to the same measurement range to the same nodes (Mz) of the common map (C), and/or maps measured values relating to adjacent measurement ranges to adjacent nodes of the common map (C) d) wherein, for a plurality of recording times (ta; ta,1, ta,2, ta,3, ta,4) or recording time intervals, events (E1, E2, E3, E4) are created based on the determined sensor measurements (m1, m2) of one or more sensors (S1, S2) according to predefined criteria, wherein, in the case of the creation of an event (E1, E2, E3, E4), the following are determined or specified: - the recording time (ta) or a recording time interval; - for each node (Nx, Ny) of the event map (C1, C2), an intensity value (I) assigned to the event (E1, E2, E3, E4); - a decay function (a1(t); a2(t); a3(t)) to define a temporal Decay behavior based on the type of event (E1, E2, E3, E4),wherein the decay function has a predetermined neutral value (0) or converges to this neutral value (0) when a predetermined time interval (ta) is exceeded, e) for the detection time (td) and for a number of nodes (Mz) of the common map (C) an event value (Vz) is determined node-wise by the individual intensity values (I) of all events (E1, E2, E3, E4) relating to those nodes (Nx, Ny) of the event map (C1, C2) which are mapped onto the respective nodes (Mz) of the common map (C) by the mapping (T1, T2) or are used for interpolation, or by weighting contributions (J) derived from these values and accumulating them with an accumulation rule in which the accumulation of the neutral value (0) makes no contribution to the accumulation result, and thus the event value (Vz) for the respective detection time (td) is determined, f) wherein the intensity values (I) or the contributions (J),which provide the intensity values (I) of events for accumulation, are each weighted with a value corresponding to the result of applying the decay function of the respective event to the time interval (td-ta) between the detection time (td) and the recording time (ta) or a time within the recording time interval of the event (E1, E2, E3, E4) when creating the event value (Vz) during the accumulation process, and g) that if an event value (Vz) in the common map (C) exceeds a predefined threshold value (Th), a critical event (EK) is detected in the relevant node (Mz) of the common map (C) and in the area assigned to this node (Mz) within the measurement range of one of the sensors (S1, S2).