Real-time leak detection correction

Abstract

In some examples, an apparatus may include a leak-detection sensor circuit, at least one hardware processor, and a memory storing machine-readable instructions that, when executed by the at least one hardware processor, cause the at least one hardware processor to receive, from the leak-detection sensor circuit, sampled leak-detection data for a potential leak. The machine-readable instructions, when executed by the at least one hardware processor, further cause the at least one hardware processor to process the sampled leak-detection data with respect to noise and signal fluctuations, determine, based on the processed sampled leak-detection data, a first derivative and a second derivative, and generate, based on the first derivative and the second derivative, an indication of whether the potential leak represents an actual leak.

Description

REAL-TIME LEAK DETECTION CORRECTIONBACKGROUND

[0001] Leak detection sensors provide for identification and addressing of gas or liquid ieaks in various types of environments such as industrial, residential, and commercial environments. Such sensors can include various operational requirements related, for example, to warm-up time, noise characteristics, and signal stabilization.BRIEF DESCRIPTION OF DRAWINGS

[0002] Features of the present disclosure are illustrated by way of example and not limited in the following figure(s), in which like numerals indicate like elements, in which:

[0003] Figure 1 illustrates a layout of a real-time leak detection correction apparatus, in accordance with an example of the present disclosure;

[0004] Figure 2 illustrates raw Analog-to-Digital Converter (ADC) samples and the corresponding smoothed output using a moving average filter, in accordance with an example of the present disclosure;

[0005] Figure 3 illustrates different voltage drops with respect to an ADC reformer, each representing a different gas flow rate leak detected during that period, in accordance with an example of the present disclosure;

[0006] Figure 4 illustrates frequency response of a filter and the time-domain representation of input and output signals, in accordance with an example of the present disclosure;

[0007] Figure 5 illustrates a normalized slope outside a dynamic threshold where it will be classified as a signal, in accordance with an example of the present disclosure;

[0008] Figure 6 illustrates an integration process involving summing up the area under the curve of the normalized first derivative over time, resulting in an ADC reformed signal that represents a cumulative sum of slope data, in accordance with an example of the present disclosure;

[0009] Figure 7 illustrates an example block diagram for real-time leak detection correction, in accordance with an example of the present disclosure;

[0010] Figure 8 illustrates a flowchart of an example method for real-time leak detection correction, in accordance with an example of the present disclosure; and

[0011] Figure 9 illustrates another example block diagram for real-time leak detection correction, in accordance with another example of the present disclosure.DETAILED DESCRIPTION

[0012] For simplicity and illustrative purposes, the present disclosure is described by referring mainly to examples. In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present disclosure. It will be readily apparent however, that the present disclosure may be practiced without limitation to these specific details. In other instances, some methods and structures have not been described in detail so as not to unnecessarily obscure the present disclosure.

[0013] Throughout the present disclosure, the terms "a" and "an" are intended to denote at least one of a particular element. As used herein, the term "includes" means includes but not limited to, the term "including" means including but not limited to. The term "based on" means based at least in part on.

[0014] Apparatuses for real-time leak detection correction, methods for real-time leak detection correction, and non-transitory computer readable media having stored thereon machine-readable instructions to provide real-time leak detection correction are disclosed herein.

[0015] With respect to leak detection generally, current leak detection firmware can be sensor specific. In this regard, when a leak detection sensor reaches its end of life (EOL), the firmware typically needs to be customized to accommodate the new sensor’s characteristics. The characteristics can include parameters such as warm-up time, noise levels, stabilization requirements, and signal integrity. This dependency on sensorspecific customization can be time-consuming and costly. In some examples, existing sensors require a relatively lengthy (e.g., 50 seconds or greater) warm-up period before they can detect leaks, further delaying response time. This process can be Inefficient andlimit the flexibility of existing leak detection sensors, resulting In suboptimal performance and high maintenance costs.

[0016] in order to address at least the aforementioned drawbacks associated with existing leak detection sensors, the apparatuses and methods disclosed herein enable virtually immediate leak detection without requiring a sensor thereof to undergo a warmup period. The apparatuses and methods disclosed herein also enhance signal integrity, reducing noise and fluctuations, leading to more accurate leak detection. Yet further, the apparatuses and methods disclosed herein provide versatility and adaptability to a variety of sensor technologies and manufacturers, eliminating the need for frequent firmware customization and improving system performance.

[0017] According to examples disclosed herein, the apparatuses and methods disclosed herein provide for sensor versatility in that the sensor firmware algorithm including a data smoother, a real-time slope corrector, and an Analog-to-Digital Converter (ADC) input reformer, supports a wide range of sensor technologies and manufacturers. This versatility ensures compatibility across different sensor types, eliminating the need for sensor-specific firmware updates.

[0018] According to examples disclosed herein, the apparatuses and methods disclosed herein provide for virtually instant detection. In this regard, the apparatuses and methods disclosed herein remove the need for sensor warm-up time, and thus provide for virtually instant leak detection to improve response times.

[0019] According to examples disclosed herein, the apparatuses and methods disclosed herein provide for enhanced signal integrity. In this regard, the apparatuses and methods disclosed herein provide for improved signal quality by correcting noise andfluctuations in real time, ensuring more accurate and reliable leak detection results.

[0020] According to exampies disclosed herein, the apparatuses and methods disclosed herein provide for elimination of warm-up time, whereby real-time leak detection correction allows for immediate leak detection, improving the overall response time of the sensor thereof.

[0021] According to examples disclosed herein, the apparatuses and methods disclosed herein provide for improved signal integrity, whereby signal quality is enhanced by reducing noise and stabilizing fluctuations, leading to more accurate detection.

[0022] According to examples disclosed herein, the apparatuses and methods disclosed herein provide for compatibility with various sensor technologies. In this regard, the firmware algorithm including the data smoother, the real-time slope corrector, and the ADC input reformer are versatile and can be adapted to work with different sensor technologies, reducing the need for costly and time-consuming firmware customizations.

[0023] According to examples disclosed herein, the apparatus may include a leakdetection sensor circuit, at least one hardware processor, and a memory storing machine-readable instructions that, when executed by the at least one hardware processor, cause the at least one hardware processor to receive, from the leak-detection sensor circuit, sampled leak-detection data for a potential leak. The machine-readable instructions, when executed by the at least one hardware processor, further cause the at least one hardware processor to process the sampled leak-detection data with respect to noise and signal fluctuations, determine, based on the processed sampled leak-detection data, a first derivative and a second derivative, and generate, based on the first derivative and the second derivative, an indication of whether the potential leak represents an actualleak.

[0024] According to exampies disclosed herein, the apparatus may include machine-readable instructions that, when executed by the at least one hardware processor, further cause the at least one hardware processor to utilize an Analog-to-Digital Converter (ADC) to sample the leak-detection data to generate the sampled leak-detection data.

[0025] According to examples disclosed herein, the apparatus may include machine-readable instructions that, when executed by the at least one hardware processor, further cause the at least one hardware processor to perform instantaneous detection of the actual leak without requiring a warm-up period.

[0026] According to examples disclosed herein, the machine-readable instructions to process the sampled leak-detection data with respect to noise and signal fluctuations, when executed by the at least one hardware processor, further cause the at least one hardware processor to utilize a moving average filter to process the sampled leakdetection data with respect to noise and signal fluctuations.

[0027] According to examples disclosed herein, the machine-readable instructions to generate, based on the first derivative and the second derivative, the indication of whether the potential leak represents the actual leak, when executed by the at least one hardware processor, further cause the at least one hardware processor to process the first derivative through an infinite impulse response (IIR) low-pass filter to generate a processed first derivative.

[0028] According to examples disclosed herein, the machine-readable instructions to generate, based on the first derivative and the second derivative, the indication of whetherthe potential leak represents the actual leak, when executed by the at least one hardware processor, further cause the at least one hardware processor to perform dynamic thresholding on the second derivative to differentiate between noise and acceptable signal levels.

[0029] According to examples disclosed herein, the machine-readable instructions to generate, based on the first derivative and the second derivative, the indication of whether the potential leak represents the actual leak, when executed by the at least one hardware processor, further cause the at least one hardware processor to determine, based on the dynamic thresholding and the processed first derivative, a normalized first derivative.

[0030] According to examples disclosed herein, the machine-readable instructions to generate, based on the first derivative and the second derivative, the indication of whether the potential leak represents the actual leak, when executed by the at least one hardware processor, further cause the at least one hardware processor to perform dynamic thresholding on the normalized first derivative, and perform, based on the dynamic thresholding performed on the normalized first derivative, signal integration and leak flow conversion to generate the indication of whether the potential leak represents the actual leak.

[0031] According to examples disclosed herein, a computer implemented method may include receiving, by at least one hardware processor, from a leak-detection sensor circuit, data for a potential leak. The method may further include processing, by the at least one hardware processor, the leak-detection data with respect to noise and signal fluctuations. The method may further include determining, by the at least one hardware processor, based on the processed leak-detection data, a first derivative and a secondderivative. The method may further include generating, by the at least one hardware processor, based on the first derivative and the second derivative, an indication of whether the potential leak represents an actual leak.

[0032] According to examples disclosed herein, a non-transitory computer readable medium having stored thereon machine-readable instructions that, when executed by at least one hardware processor, cause the at least one hardware processor to receive, from a leak-detection sensor circuit, leak-detection data for a potential leak. The machine-readable instructions, when executed by at least one hardware processor, further cause the at least one hardware processor to determine based on the leak-detection data, a first derivative and a second derivative, and generate based on the first derivative and the second derivative, an indication that the potential leak represents an actual leak. The machine-readable instructions, when executed by at least one hardware processor, further cause the at least one hardware processor to control, based on the indication that the potential leak represents the actual leak, operation of a system associated with the actual leak.

[0033] For the apparatuses, methods, and non-transitory computer readable media disclosed herein, the elements of the apparatuses, methods, and non-transitory computer readable media disclosed herein may be any combination of hardware and programming to implement the functionalities of the respective elements. In some examples described herein, the combinations of hardware and programming may be implemented in a number of different ways. For example, the programming for the elements may be processor executable instructions stored on a non-transitory machine-readable storage medium and the hardware for the elements may Include a processing resource to execute thoseinstructions. In these examples, a computing device implementing such elements may include the machine-readable storage medium storing the instructions and the processing resource to execute the instructions, or the machine-readable storage medium may be separately stored and accessible by the computing device and the processing resource. In some examples, some elements may be implemented in circuitry.

[0034] Figure 1 illustrates a layout of an example real-time leak detection correction apparatus (hereinafter also referred to as “apparatus 100”).

[0035] Referring to Figure 1, the apparatus 100 may include a leak-detection sensor circuit 102.

[0036] The apparatus 100 may further include at least one hardware processor (e.g., the hardware processor 702 of Figure 7, and / or the hardware processor 904 of Figure 9), and a memory (e.g., the memory 704 of Figure 7) storing machine-readable instructions that, when executed by the at least one hardware processor, cause the at least one hardware processor to receive, from the leak-detection sensor circuit 102, sampled leakdetection data (e.g., data from output 104 that is sampled by an Analog-to-Digital Converter (ADC) 106) for a potential leak.

[0037] The machine-readable instructions, when executed by the at least one hardware processor, further cause the at least one hardware processor to process (e.g., by a data smoother 108) the sampled leak-detection data with respect to noise and signal fluctuations.

[0038] The machine-readable instructions, when executed by the at least one hardware processor, further cause the at least one hardware processor to determine,based on the processed sampled leak-detection data, a first derivative 110 and a second derivative 112.

[0039] The machine-readable instructions, when executed by the at least one hardware processor, further cause the at least one hardware processor to generate, based on the first derivative 110 and the second derivative 112, an indication (e.g., by a leak flow indicator 130) of whether the potential leak represents an actual teak. The indication of whether the potential teak represents an actual teak may be displayed on a display 136.

[0040] According to examples disclosed herein, the machine-readable instructions, when executed by at least one hardware processor, further cause the at least one hardware processor to control, based on the indication that the potential leak represents the actual leak, operation of a system 138 associated with the actual leak. In this regard, the output generated by a leak detector as disclosed herein may be utilized in an automation system, for example, through USB communication. In one example, the automation system as disclosed herein may monitor health of the apparatus or another machine to prevent hazards or abnormalities.

[0041] According to examples disclosed herein, the apparatus 100 may include machine-readable instructions that, when executed by the at least one hardware processor, further cause the at least one hardware processor to utilize the ADC 106 to sample the leak-detection data to generate the sampled leak-detection data.

[0042] According to examples disclosed herein, the machine-readable instructions to process the sampled leak-detection data with respect to noise and signal fluctuations, when executed by the at least one hardware processor, further cause the at least onehardware processor to utilize a moving average filter 114 to process the sampled leakdetection data with respect to noise and signal fluctuations.

[0043] According to examples disclosed herein, the machine-readable instructions to generate, based on the first derivative 110 and the second derivative 112, the indication of whether the potential leak represents the actual leak, when executed by the at least one hardware processor, further cause the at least one hardware processor to process the first derivative 110 through an infinite impulse response (IIR) low-pass filter 116 to generate a processed first derivative.

[0044] According to examples disclosed herein, the machine-readable instructions to generate, based on the first derivative 110 and the second derivative 112, the indication of whether the potential leak represents the actual leak, when executed by the at least one hardware processor, further cause the at least one hardware processor to perform (e.g., by a real-time slope corrector 118) dynamic thresholding 120 on the second derivative 112 to differentiate between noise and acceptable signal levels.

[0045] According to examples disclosed herein, the machine-readable instructions to generate, based on the first derivative 110 and the second derivative 112, the indication of whether the potential leak represents the actual leak, when executed by the at least one hardware processor, further cause the at least one hardware processor to determine, based on the dynamic thresholding 120 and the processed first derivative, a normalized first derivative 122.

[0046] According to examples disclosed herein, the machine-readable instructions to generate, based on the first derivative 110 and the second derivative 112, the indication of whether the potential leak represents the actual leak, when executed by the at leastone hardware processor, further cause the at least one hardware processor to perform dynamic thresholding 140 on the normalized first derivative, and perform, based on the dynamic thresholding performed on the normalized first derivative, (e.g., by ADC input reformer 124) signal integration and voltage decay compensation 126 and leak flow conversion 128 to generate the indication (e.g., by the leak flow indicator 130) of whether the potential leak represents the actual leak.

[0047] With continued reference to Figure 1, with respect to data sampling and filtering, an output 104 of the sensor circuit 102 may be passed through a low-pass filter 134 and periodically sampled by the ADC 106. Given an ADC samples per second (SPS) also known as sampling frequency (Fs), the sampling period (Ts) may be determined as follows:1~ ““sFThe output 104 may be sampled via, for example, an Inter-Integrated Circuit (I2C) communication interface.

[0048] The data smoother 108 may process the ADC 106 samples using a moving average filter 114 to smooth out noise and signal fluctuations. Smoothing involves averaging a set number of sequential data points to produce a smoothed output. The moving average filter may be represented as follows:Here (MA[n]) is the smoothed output at the (n)-th sample, (x[n]) is the input ADC sample at the (n)-th sample, and (N) is the number of samples in the moving average window.The window size (N) significantly affects the signal response time. A larger window size results in a smoother output but also causes the response to become slower. Conversely, a smaller window size provides a faster response but may not effectively smooth out noise and fluctuations. The sample lag introduced by the moving average filter can be determined as follows:__ (NMA- 1)LagMA-- -To ensure a system’s response time remains within the required limits, the maximum window size (NMA_max) can be determined by maximum signal lag in the system (T_lag):This equation ensures that the moving average window size does not exceed the specified time lag, maintaining the system’s responsiveness. By effectively reducing the impact of transient noise and fluctuations in the signal, this technique results in a more stable and reliable output. Figure 2 illustrates raw ADC samples and the corresponding smoothed output using a moving average filter, in accordance with an example of the present disclosure.

[0049] The smoothed data may then be used to determine a first derivative 110 and a second derivative 112 to assess changes over time. With respect to determination of the first derivative 110 and the second derivative 112, the smoothed data points may be processed through the central difference equation to obtain the rate of change at each point, which helps identifying trends and patterns in the data overtime as follows:Here (MA[x]) is the smoothed output at the current sample, (MA[x-1]) is the smoothed output at the previous sample, and (T__s) is the sampling period. The second derivative provides insight into the acceleration or deceleration of the signal, and can compensate the drift.

[0050] The first derivative 110, also known as the slope, may be determined to compensate for drift caused by heat from the sensor circuit 102. In this regard, as the sensor circuit 102 heats up, the output signal (e.g., the output 104) may drift, leading to inaccuracies.

[0051] Referring to Figures 2-5, during startup, the sensor voltage exponentially drifts as shown at 304 in Figure 3. In this regard, Figure 2 shows signal detection when the voltage rapidly drops. For example, a different moving average configuration at 200, 202, and 204 will contribute lag to the signal response time. For example, the signal at 204 with a moving average 25 will be late by 12 samples. After performing the first derivative, the initial part of the first derivative will also show exponential drift as shown at location 400 of Figure 4). To detect the signal without being affected by the drift data, the second derivative can be used as shown at location 500 of Figure 5). By checking the second derivative data, it is possible to detect the signal’s rising (e.g., at 300 of Figure 3) or falling edge (e.g., at 302 to Figure 3).

[0052] The first derivative 110 may be processed through an infinite impulse response (IIR) low-pass filter 116, which reduces high-frequency noise while preserving lower-frequency signal components. In one example, the IIR low-pass filter employed mayinclude a second-order Butterworth filter with a specified cutoff frequency (Fc). This filter may be implemented using a series of biquad sections arranged in a Direct Form II Transposed structure, chosen for its numerical stability and efficient computational performance. The cutoff frequency (Fc) may be selected based on the desired frequency range of the signal components to be preserved. For example, if the signal of interest lies below 1 kHz, the cutoff frequency (Fc) may be set to 1 kHz. The input signal may be sampled at a frequency (Fs), which may be at least twice the highest frequency component of the signal to satisfy the Nyquist criterion. This ensures that the signal is adequately sampled without aliasing.

[0053] The second-order Butterworth filter may be implemented using biquad sections. Each biquad section may process the signal through a pair of second-order filter stages. The Direct Form II Transposed structure may be utilized for its advantages in numerical stability and reduced sensitivity to coefficient quantization errors. The coefficients for the Butterworth filter may be calculated to ensure a maximally flat frequency response in the passband. These coefficients may be derived from Butterworth polynomial equations, which define the filter’s transfer function. The coefficients may then be used to configure the biquad sections.

[0054] Referring to Figure 4, the input signal (f’[n]) (e.g., at 400 of Figure 4), which contains both the desired low-frequency components and high-frequency noise, may be fed into the filter. The output signal (y’[n]) (e.g., at 402 of Figure 4) may be obtained by passing the input signal through the IIR low-pass filter. The filter attenuates frequencies above the cutoff frequency (Fc), effectively reducing high-frequency noise while preserving the low-frequency components of the signal. The result is a cleaner signalwith minimal distortion in the passband.

[0055] To illustrate the effect of the low-pass filter, Figure 4 illustrates the frequency response of the filter and the time-domain representation of the input (e.g., at 400 of Figure 4) and output signals (e.g., at 402 of Figure 4).

[0056] The input signal (f’[n]) and output signal (y’[n]) variables may be defined as follows:y'[n] = b0f'[n] + d1d2= b2f'[n] + a2y'[n]Where:d1, d2are state variableb0, b1, b2are feedforward coefficients,a0, a1are feedback coefficients

[0057] Next, with respect to real-time slope correction, the real-time slope corrector 118 accounts for sensor variations, including differences in batches, models, and manufacturers, which result in distinct heating profiles. These variations may cause the output voltage of the sensor circuit 102 to drift over time. The real-time slope corrector 118 may adjust the drifted slope by aligning it with a pre-established baseline slope.

[0058] Referring to Figures 1 and 4, with respect to the difference heating profile, the slope has a different startup level (e.g., at 400 of Figure 4), and thus will take a longer time to reach steady state. The real-time slope corrector 118 may help to adjust the slopefrom a drifted state (e.g., at 402 of Figure 4) into a normalized state (e.g., at 404 of Figure 4).

[0059] Referring to Figures 1 and 3, this correction may center an output of the sensor circuit 102 to a standard reference point, ensuring consistent performance and reducing the settling time. In this regard, with respect to the DC drift 304 and after restructure the DCV value 306 (e.g., the same as the output of the signal integration and voltage decay compensation 126), since the DC voltage is at a normalized level, a user may immediately perform the detection. In this regard, comparison detection between raw voltage and normalized voltage (e.g., reformed DC) for level 1 detection shows that the startup leakage detected voltage drop at 308 is larger than at 310, even though both have the same leakage rate. This discrepancy can cause the system to report incorrect and unstable leakage detection. Using raw voltage as the detection level requires an increased warm-up time until the raw voltage stabilizes and exhibits low drift over time. With the normalized DC level at 312 and 314, the signal voltage drop level is maintained, allowing the system to perform immediate detection. This real-time correction improves the accuracy of gas leakage detection, particularly during the early stages of detection.

[0060] Referring to Figure 1, with respect to dynamic thresholding and baseline detection, the real-time slope corrector 118 may perform dynamic thresholding 120 to differentiate between noise and true signal levels. The dynamic thresholding 120 operates within a window period of (X) seconds, during which data is collected to analyze the noise level. With respect to dynamic thresholding, during the window period, data points may be collected to establish a noise profile. The collected data may then be analyzed to determine the noise level by calculating statistical measures such as the mean andstandard deviation. Based on this analysis, a dynamic threshold is set, which adapts to changes in noise levels over time. This ensures that only significant signal variations are detected. Data points that exceed the dynamic threshold are classified as true signals, while those below are considered noise.

[0061] Referring to Figure 5, during the startup process, the second derivative 500 and normalized slope 502 data are initially classified as baseline. This initial data is used to determine the noise level.

[0062] Referring to Figure 5, after the data collection is complete, only data within the threshold level is classified as baseline, and current data (e.g., at 500 and 502) will proceed for further noise analysis as input for dynamic threshold. Thus, data classified as baseline may pass through a dynamic thresholding window. In this regard, the normalized slope (e.g., at 502) and dynamic thresholding for the normalized slope (e.g., at 504) can be illustrated, as well as the 2ndderivatives result (e.g., at 500) and dynamic thresholding for the 2ndderivatives (e.g., at 506).

[0063] The dynamic thresholding 120 may utilize a moving standard deviation technique and apply, for example, a 3-sigma rule or a gain multiplier rule to establish the noise threshold. The moving standard deviation technique Involves determining the standard deviation (o) of the data points within the window period. The moving standard deviation may be determined as follows:Here, (K) is the window size, (xj) represents each individual data point, and (p) is themean of the data points. The noise threshold is then established by applying a gain multiplier to the standard deviation. The threshold may be determined as follows:threshold ~ gain x aTo define the gain, increasing the gain setting will result in a higher confidence level to consider the data as baseline. In one example, a standard gain value of 2 may be utilized. The range for the gain may be set, for example, between 2 and 6. By adjusting the gain within this range, the threshold can be fine-tuned to effectively distinguish between noise and true signals.

[0064] Both the normalized first derivative 122 and the second derivative 112 may be evaluated against this dynamic threshold for noise level (e.g., at 120 and 140). In this regard, during startup, due to sensor warm-up, the first derivatives may exhibit an exponential curve (e.g., at 502), causing the second derivatives to slightly fluctuate (e.g., at 500), and considered as noise. The second derivative dynamic threshold will react accordingly (e.g., at 506). After some period, once the second derivative becomes stable (e.g., at 508), the dynamic threshold (e.g., at 510) will adjust accordingly.

[0065] Data points where the first and second derivatives are below the dynamic threshold may be considered baseline, and the drift in the first derivative 110 may be corrected in real-time by the baseline slope. In this regard, for the absolute value of the second derivative signal that is above the second derivative threshold (e.g., at 512), it is considered a signal. There is also a signal case where the second derivative is within the second derivative threshold (e.g., at 514). In this case, the normalized slope (e.g., at 516) may need to be monitored. If the normalized slope is outside the slope threshold, it is considered a signal. Otherwise, all conditions will be considered as baseline.

[0066] The equations for data collection and ignoring signal data may be specified as follows:Baseline ~ {x / ’W) < 7’i and f"(xj < T2Here, (xj) represents an individual data point in the time series, (f’(xj)) denotes the first derivative of the data point (xj), and (f”(xj)) denotes the second derivative of the data point (xj). The parameters (T_1) and (T_2) are the dynamic thresholds for the first and second derivatives, respectively.Here, data points where either the first derivative or the second derivative exceeds their respective thresholds are considered signals.Collected Data ~ fWfi 6 Baseline]Here, only data points classified as baseline are collected for dynamic thresholding moving standard deviation and slope baseline moving average.Here, data points classified as signals are ignored in the data collection process.

[0067] All baseline data points may be further smoothed using a moving average filter to establish a stable slope baseline (e.g., a 1stderivative baseline 132). Smoothing may be performed as follows:Slope baseline = {fWf'i 6 Baseline]Here, only first derivative points classified as baseline are collected for slope baseline moving average. The output of the slope moving average will consider as slope baseline.Referring to Figure 4, the slope baseline remain flat (e.g., at 406) in signal state.

[0068] A normalized first derivative 110 (e.g., with drift removed) may be obtained by subtracting the slope baseline from the first derivative data as follows:, Steps average sredin this regard, for the example of Figures 4 and 5, the normalized first derivative is shown at 502, the current first derivative is shown at 402, and the slope baseline is shown at 406.

[0069] Once the normalized first derivative 110 has been determined, the ADC input reformer 124 may perform leak flow conversion 128 to convert slope data into direct current (DC) data by first performing signal integration and voltage decay compensation 126 to integrate the normalized first derivative 110. The ADC input reformer 124 facilitates correction of the drift DC signal into normalized DC data. By using normalized DC data, the ADC input reformer 124 performs the calibration for the leak rate versus DC value.

[0070] All normalized first derivative inputs that exceed the dynamic threshold may be classified as a signal. In this regard, when the normalized first derivative inputs exceed the dynamic threshold, they are classified as a signal. For example, Figure 5 shows that since the normalized slope (e.g., at 518) is outside the dynamic threshold (e.g., at 520), it will be classified as a signal. Similarly, if the second derivative (e.g., at 522) is outside the dynamic threshold (e.g., at 512), it will also be classified as a signal.

[0071] When the apparatus 100 is brought to a gas leak source, the normalized first derivative curve exceeds the threshold, indicating the presence of a gas leak. At this point, integration is performed from the normalized slope to convert this data into a DC signal for further analysis and calibration.

[0072] In order to obtain a signal voltage, the data is then integrated back into DC voltage. In this regard, Figure 6 shows the integration process involves summing up the area under the curve (e.g., at 600 and 602) of the normalized first derivative over time, resulting in an ADC reformed signal 604 that represents the cumulative sum of the slope data.

[0073] The signal DC value, as shown below, for the leak flow indicator 130 may be determined by identifying and subtracting the slope baseline, which represents voltage decay over time. In this regard, referring to Figures 1 and 3, the leak flow indicator 130 may be calibrated based on the ADC reformer output 306. For example, the ADC reformer output 306 shows seven different voltage drops, each representing a different gas flow rate leak detected during that period. A larger ADC reformer output indicates a significant gas leak.

[0074] This process compensates for sensor drift and ensures that the final DC value reflects true signal activity by compensating for any potential decline in sensor performance over time. The signal DC level may be determined by summing the area under the normalized first derivative. The signal DC level may be determined as the sum of the area under normalized first derivative as follows:Here, (f’„nor(t)) represents the normalized first derivative of the signal, (V__reform) represents the signal DC level, representing the cumulative sum of the slope data over time, and (tJD) and (t__1) represent the time points where the slope exceeds the dynamic threshold.

[0075] Factory calibration processes can be conducted at various levels (1, 3, and 5) to ensure the system’s accuracy in different operating environments. These levels represent different leak rates, which are related by an exponential curve. For example, Level 1 corresponds to the minimum leak rate, also known as the sensitivity of the device. Levels 3 and 5 represent higher leak rates, with each level capturing a specific range of leak rates.

[0076] The calibration system may capture the correlation between the corrected DC level and the gas leak flow rate 128. This corrected DC value may then be stored inside the EEPROM as a bar level specified by the leak detection indicator 130.

[0077] Figures 7-9 respectively illustrate an example block diagram 700, a flowchart of an example method 800, and a further example block diagram 900 for real-time leak detection correction, according to examples. The block diagram 700, the method 800, and the block diagram 900 may be implemented on the apparatus 100 described above with reference to Figure 1 by way of example and not of limitation. The block diagram 700, the method 800, and the block diagram 900 may be practiced in other apparatus. In addition to showing the block diagram 700, Figure 7 shows hardware of the apparatus 100 that may execute the instructions of the block diagram 700. The hardware may include a processor 702, and a memory 704 storing machine-readable instructions that when executed by the processor cause the processor to perform the instructions of theblock diagram 700. The memory 704 may represent a non-transitory computer readable medium. Figure 8 may represent an example method for real-time leak detection correction, and the steps of the method. Figure 9 may represent a non-transitory computer readable medium 902 having stored thereon machine-readable instructions to perform real-time leak detection correction, according to an example. The machine-readable instructions, when executed, cause a processor 904 to perform the instructions of the block diagram 900 also shown in Figure 9.

[0078] The processor 702 of Figure 7, and the processor 904 of Figure 9 may include a single or multiple processors or other hardware processing circuit, to execute the methods, functions and other processes described herein. These methods, functions and other processes may be embodied as machine-readable instructions stored on a computer readable medium, which may be non-transitory (e.g., the non-transitory computer readable medium 902 of Figure 9), such as hardware storage devices (e.g., RAM (random access memory), ROM (read only memory), EPROM (erasable, programmable ROM), EEPROM (electrically erasable, programmable ROM), hard drives, and flash memory). The memory 704 may include a RAM, where the machine-readable instructions and data for a processor may reside during runtime.

[0079] Referring to Figures 1-7, and particularly to the block diagram 700 shown in Figure 7, the memory 704 may include instructions 706 to receive, from the leak-detection sensor circuit 102, sampled leak-detection data (e.g., data from output 104 that is sampled by an Analog-to-Digital Converter (ADC) 106) fora potential leak.

[0080] The processor 702 may fetch, decode, and execute the instructions 708 to process (e.g., by the data smoother 108) the sampled leak-detection data with respect tonoise and signal fluctuations.

[0081] The processor 702 may fetch, decode, and execute the instructions 710 to determine, based on the processed sampled leak-detection data, a first derivative 110 and a second derivative 112.

[0082] The processor 702 may fetch, decode, and execute the instructions 712 to generate, based on the first derivative 110 and the second derivative 112, an indication (e.g., by a leak flow indicator 130) of whether the potential leak represents an actual leak.

[0083] Referring to Figures 1-6 and 8, and particularly Figure 8, for the method 800, at block 802, the method may include receiving, from the leak-detection sensor circuit 102, data for a potential leak.

[0084] At block 804, the method may include processing (e.g., by the data smoother 108) the leak-detection data with respect to noise and signal fluctuations.

[0085] At block 806, the method may include determining, based on the processed leak-detection data, a first derivative 110 and a second derivative 112.

[0086] At block 808, the method may include generating, based on the first derivative 110 and the second derivative 112, an indication (e.g., by a leak flow indicator 130) of whether the potential leak represents an actual leak.

[0087] Referring to Figures 1 and 9, and particularly Figure 9, for the block diagram 900, the non-transitory computer readable medium 902 may include instructions 906 to receive, from a leak-detection sensor circuit 102, leak-detection data for a potential leak.

[0088] The processor 904 may include instructions 908 to determine, based on the leak-detection data, a first derivative 110 and a second derivative 112.

[0089] The processor 904 may include instructions 910 to generate, based on the first derivative 110 and the second derivative 112, an indication (e.g., by a leak flow indicator 130) that the potential leak represents an actual leak.

[0090] The processor 904 may include instructions 912 to control, based on the indication that the potential leak represents the actual leak, operation of a system 138 associated with the actual leak. The output generated by the apparatus 100 may be utilized in an automation system (e.g., the system 138), for example, through USB communication. In one example, the automation system may monitor health of a machine to prevent hazards or abnormalities. Examples of such machines / instruments may include gas chromatographs (GCO, gas chromatograph-mass spectrometers (GC / MS), inductively coupled plasma optical emission spectrometers (ICP-OES), ICP mass spectrometers (ICP-MS), etc. With respect to such instruments, leaks may lead to shortened lifetimes of some consumables, especially in GC systems. Further, leaks allow gas to flow out of a fitting and can cause loss of production and revenue, if trying to produce or collect and supply gas. Yet further, if gas is being consumed at a higher rate than expected because of leakage, gas supply costs may likely increase. Checking for and eliminating leaks from a gas source to the terminus can reduce safety concerns and reduce costs through decreased use of cylinders or fewer refills by a gas supply company. Regular leak checks of gas tubing fittings and regulators with the apparatus 100 may thus keep leaks at a minimum and maintain optimal conditions for a laboratory or site.

[0091] What has been described and illustrated herein is an example along with some of its variations. The terms, descriptions and figures used herein are set forth by way of illustration only and are not meant as limitations. Many variations are possible within thespirit and scope of the subject matter, which is intended to be defined by the following claims -and their equivalents -in which ail terms are meant in their broadest reasonable sense unless otherwise indicated.

Claims

What is claimed is:

1. An apparatus comprising:a leak-detection sensor circuit;at least one hardware processor; anda memory storing machine-readable instructions that, when executed by the at least one hardware processor, cause the at least one hardware processor to:receive, from the leak-detection sensor circuit, sampled leak-detection data for a potential leak;process the sampled leak-detection data with respect to noise and signal fluctuations;determine, based on the processed sampled leak-detection data, a first derivative and a second derivative; andgenerate, based on the first derivative and the second derivative, an indication of whether the potential leak represents an actual leak.

2. The apparatus according to claim 1, further comprising machine-readable instructions that, when executed by the at least one hardware processor, further cause the at least one hardware processor to:utilize an Analog-to-Digital Converter (ADC) to sample the leak-detection data to generate the sampled leak-detection data.

3. The apparatus according to claim 1, wherein the machine-readable instructions to process the sampled leak-detection data with respect to noise and signal fluctuations, when executed by the at least one hardware processor, further cause the at least one hardware processor to:utilize a moving average filter to process the sampled leak-detection data with respect to noise and signal fluctuations.

4. The apparatus according to claim 1, wherein the machine-readable instructions to generate, based on the first derivative and the second derivative, the indication of whether the potential leak represents the actual leak, when executed by the at least one hardware processor, further cause the at least one hardware processor to:process the first derivative through an infinite impulse response (IIR) low-pass filter to generate a processed first derivative.

5. The apparatus according to claim 4, wherein the machine-readable instructions to generate, based on the first derivative and the second derivative, the indication of whether the potential leak represents the actual leak, when executed by the at least one hardware processor, further cause the at least one hardware processor to:perform dynamic thresholding on the second derivative to differentiate between noise and acceptable signal levels.

6. The apparatus according to claim 5, wherein the machine-readable instructions to generate, based on the first derivative and the second derivative, the indication of whether the potential leak represents the actual leak, when executed by the at least one hardware processor, further cause the at least one hardware processor to:determine, based on the dynamic thresholding and the processed first derivative, a normalized first derivative.

7. The apparatus according to claim 6, wherein the machine-readable instructions to generate, based on the first derivative and the second derivative, the indication of whether the potential leak represents the actual leak, when executed by the at least one hardware processor, further cause the at least one hardware processor to:perform dynamic thresholding on the normalized first derivative: andperform, based on the dynamic thresholding performed on the normalized first derivative, signal integration and leak flow conversion to generate the indication of whether the potential leak represents the actual leak.

8. A computer implemented method comprising:receiving, by at least one hardware processor, from a leak-detection sensor circuit, data for a potential leak;processing, by the at least one hardware processor, the data with respect to noise and signal fluctuations;determining, by the at least one hardware processor, based on the processed data, a first derivative and a second derivative; andgenerating, by the at least one hardware processor, based on the first derivative and the second derivative, an indication of whether the potential leak represents an actual leak.

9. The method according to claim 8, further comprising:utilizing an Analog-to-Digital Converter (ADC) to sample the data to generate sampled leak-detection data.

10. The method according to claim 8, wherein processing, by the at least one hardware processor, the data with respect to noise and signal fluctuations, further comprises: utilizing, by the at least one hardware processor, a moving average filter to process the leak-detection data with respect to noise and signal fluctuations.

11. The method according to claim 8, wherein generating, by the at least one hardware processor, based on the first derivative and the second derivative, the indication of whether the potential leak represents the actual leak, further comprises:processing, by the at least one hardware processor, the first derivative through an infinite impulse response (IIR) low-pass filter to generate a processed first derivative.

12. The method according to claim 11, wherein generating, by the at least one hardware processor, based on the first derivative and the second derivative, the indication of whether the potential leak represents the actual leak, further comprises:performing, by the at least one hardware processor, dynamic thresholding on the second derivative to differentiate between noise and acceptable signal levels.

13. The method according to claim 12, wherein generating, by the at least one hardware processor, based on the first derivative and the second derivative, the indication of whether the potential leak represents the actual leak, further comprises:determining, by the at least one hardware processor, based on the dynamic thresholding and the processed first derivative, a normalized first derivative.

14. The method according to claim 13, wherein generating, by the at least one hardware processor, based on the first derivative and the second derivative, the indication of whether the potential leak represents the actual leak, further comprises:performing, by the at least one hardware processor, dynamic thresholding on the normalized first derivative; andperforming, by the at least one hardware processor, based on the dynamic thresholding performed on the normalized first derivative, signal integration and leak flow conversion to generate the indication of whether the potential leak represents the actual leak.

15. A non-transitory computer readable medium having stored thereon machine-readable instructions that, when executed by at least one hardware processor, cause the at least one hardware processor to:receive, from a leak-detection sensor circuit, leak-detection data for a potential leak; determine based on the leak-detection data, a first derivative and a second derivative;generate based on the first derivative and the second derivative, an indication that the potential leak represents an actual leak; andcontrol, based on the indication that the potential leak represents the actual leak, operation of a system associated with the actual leak.

16. The non-transitory computer readable medium according to claim 15, wherein the machine-readable instructions, when executed by the at least one hardware processor, further cause the at least one hardware processor to:perform instantaneous detection of the actual leak without requiring a warm-up period.

17. The non-transitory computer readable medium according to claim 15, wherein the machine-readable instructions to generate based on the first derivative and the second derivative, the indication that the potential leak represents the actual leak, when executed by the at least one hardware processor, further cause the at least one hardware processor to:process the first derivative through an infinite impulse response (IIR) low-pass filter to generate a processed first derivative.

18. The non-transitory computer readable medium according to claim 17, wherein the machine-readable instructions to generate based on the first derivative and the second derivative, the indication that the potential leak represents the actual leak, when executed by the at least one hardware processor, further cause the at least one hardware processor to:perform dynamic thresholding on the second derivative to differentiate between noise and acceptable signal levels.

19. The non-transitory computer readable medium according to claim 18, wherein the machine-readable instructions to generate based on the first derivative and the second derivative, the indication that the potential leak represents the actual leak, when executed by the at least one hardware processor, further cause the at least one hardware processor to:determine, based on the dynamic thresholding and the processed first derivative, a normalized first derivative.

20. The non-transitory computer readable medium according to claim 19, wherein the machine-readable instructions to generate based on the first derivative and the second derivative, the indication that the potential leak represents the actual leak, when executedby the at least one hardware processor, further cause the at least one hardware processor to:perform dynamic thresholding on the normalized first derivative; andperform, based on the dynamic thresholding performed on the normalized first derivative, signal integration and leak flow conversion to generate the indication that the potential leak represents the actual leak.

Images