Method and system for detecting operation mode of overlapping load
By measuring and analyzing the aggregated power signals of power sockets, combining the spectrum coherence of the characteristic signals of load categories, the low power mode of overlapping electrical loads is solved, and the problem of difficulty in accurately distinguishing low power modes in the prior art is improved. Accuracy and security.
Patent Information
- Application Number
- CN202280094052.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Priority Date
- 2022-03-24
- Filing Date
- 2022-06-10
- Publication Date
- 2024-11-08
AI Technical Summary
The prior art is difficult to accurately detect operating modes of multiple overlapping electrical loads, especially distinguishing low power modes in aggregate signals, resulting in insufficient energy management, which may lead to unnecessary energy consumption and equipment damage.
By measuring the aggregated power signal of the power socket, determining its operating mode, and evaluating the spectral coherence with the characteristic signal of the load category, the coherence probability is calculated using the sigmoid membership function to detect the low-power mode in the active mode, and then determining each Operating mode of the electric load.
Improves monitoring accuracy of overlapping electrical load operation modes, enhances the effectiveness of energy management, reduces unnecessary equipment shutdowns and potential damage, and provides more accurate energy saving recommendations.
Smart Images

Figure CN118922726A_ABST
Abstract
Description
Technical Field
[0001] The field of the present disclosure relates to determining an operating mode for multiple overlapping electrical loads. The present disclosure provides, among other things, a method and system for determining a low power mode within an aggregate higher power signal. Background Art
[0002] The monitoring of all power consumption of electrical loads associated with power management systems is often overlooked. In particular, for the purpose of energy management, only high-power electrical loads are usually evaluated, and for circuits to which only one load is connected. Therefore, only a single operating mode is determined. However, in a circuit powered by a circuit breaker or a power strip, there may be multiple electrical loads powered by a single socket or circuit breaker. These multiple electrical loads connected to the same circuit can operate in different operating modes at the same time. Therefore, when operating in different modes, the resulting electrical signal at the circuit breaker or power strip includes aggregated electrical signals of overlapping loads with different power. In this way, some loads may be in active mode (i.e., performing actual operations) and some loads may be in low-power mode (i.e., standby mode).
[0003] For example, electrical loads and / or appliances used in a building experience various operating modes depending on the actual work done by the load, which may also depend on the user's usage. For example, when a microwave oven is actively used for heating, it can be classified as being in active mode, and when the microwave oven is plugged into a power outlet and not used for heating, it can be classified as being in standby mode or low power mode. Through existing methods, the operating mode of the socket / load can be detected when the load is powered or operated alone (U.S. Patent No. 10837989, 2020). However, as with the example of the microwave oven, when a circuit breaker is connected to multiple kitchen appliances, not all loads will be powered at the same time.
[0004] In order to provide more accurate energy saving suggestions or power control to the end user, it is advantageous to know whether any load in the combined electrical signal is operating in a low power mode. Providing a detailed analysis of the changing operating modes (especially low power modes) of the loads from the aggregated signal gives the end user the opportunity to further save energy. After identifying the loads in low power operating mode, the user can choose to turn off these loads, thereby saving power and energy. Summary of the invention
[0005] The present invention provides a method of detecting the presence of a low power operating mode in the presence of overlapping loads. By effectively detecting the operating mode of an appliance, various benefits can be provided. It enables the customer to see the load energy / power usage by operating mode. In a networked plug-in load system, this information is particularly important for providing an aggregate energy / power consumption that differentiates / breaks down between the energy actually used and the energy the user believes is being consumed. This information allows the customer to directly see potential power saving opportunities.
[0006] In addition, reliable operation mode detection also ensures reliable control (or safe shutdown) of plug-in critical loads, minimizing potential damage to equipment and negative impact on users. Therefore, this helps minimize unnecessary nuisance trips when plug-in loads (especially critical loads) are to be shut down.
[0007] In one aspect of the present disclosure, a method for detecting an operating mode of an overlapping electrical load is provided. The method for detecting an operating mode of an overlapping electrical load comprises: measuring an aggregate power signal of an electrical outlet; determining an operating mode of the aggregate power signal, wherein the operating mode is determined to be one of an active mode, a lower power mode, and an off mode; when it is determined that the operating mode is an active mode, the method further comprises: determining a load category of the aggregate power signal; selecting a corresponding load category characteristic power signal from a load category database; evaluating the spectral coherence between the aggregate power signal and the load category characteristic power signal; determining the overall coherence probability between the frequency components of the aggregate power signal and the frequency components of the load category characteristic power signal; determining whether a low power mode exists within the aggregate power signal of the active mode, whereby the low power mode is arranged to be detectable based on the value of the coherence probability; and determining the operating mode of each of the overlapping electrical loads.
[0008] Understanding the operating patterns of overlapping electrical loads can improve power consumption monitoring and enhance energy management capabilities including effective control of the loads.
[0009] In some embodiments, measuring the aggregate power signal can include reading the aggregate current signal and the aggregate voltage signal of the power outlet. Measuring from the power outlet allows monitoring and acquisition to be performed at a single source rather than at each load. This is also advantageous if access to certain locations is limited.
[0010] In some embodiments, reading the aggregate current signal and the aggregate voltage signal may be performed for at least 5 cycles. Collecting signals for at least 5 cycles can minimize any anomalies or abnormalities caused in one cycle in the resulting measured aggregate signal.
[0011] In some embodiments, the load category of the aggregate power signal can be determined to be one of a plurality of load categories, which can include: power electronic load without power factor correction (NP); power electronic load with power factor correction (P); transformer (T); reactive (X); phase angle controlled (PAC); complex (M); or resistive (R). Knowing the load category helps determine the appliance associated with the load so that power management of such appliance can be controlled. Those skilled in the art will recognize that other load categories can be used and that the method is not limited to the load categories described above.
[0012] In some embodiments, the method may further include normalizing the aggregate power signal, wherein the aggregated current and voltage waveforms are normalized. The normalized aggregate signal can be more easily compared to the normalized sample signals because the behavior of the waveform is of interest rather than the maximum current or voltage.
[0013] In some embodiments, the load category signature power signal may include a normalized waveform in active mode for one cycle for each load category. The load category signature power signal is a sampled signal that has been calibrated for a specific load category, so only a single cycle is required when the specific load is operating in active mode.
[0014] In some embodiments, evaluating the spectral coherence between the aggregate power signal and the load category characteristic power signal may also include: evaluating the difference between the aggregate power signal and the load category characteristic power signal based on their respective voltage / current, VI trajectory waveform, and area enclosed in the VI trajectory.
[0015] In some embodiments, evaluating the spectral coherence between the aggregate power signal and the load category characteristic power signal may further include: calculating the difference between the area enclosed by the aggregate power signal and the VI trajectory of the load category characteristic power signal using a root mean square error (RMSE) analysis. The difference between the area enclosed by the aggregate power signal and the VI trajectory of the load category characteristic power (sample) signal is used as a distinguishing feature because the distinguishing feature helps to provide easier and more accurate prediction of low power loads present in the aggregate signal.
[0016] In some embodiments, evaluating the spectral coherence between the aggregate power signal and the load category characteristic power signal may also include: using the Welch mean modified periodogram method (https: / / en.wikipedia.org / wiki / Welch%27s_method) to estimate the spectral coherence between the normalized current signal of the aggregate power signal and the normalized current signal of the load category characteristic power signal. Again, this is a distinguishing feature that helps provide easier and more accurate prediction of low power loads present in the aggregate signal.
[0017] In some embodiments, evaluating the spectral coherence between the aggregate power signal and the load class characteristic power signal may further include recording an estimate of the spectral coherence of all frequency components of the compared signals.
[0018] In some embodiments, the method may further include evaluating a spectral correlation count between the aggregate power signal and the load category characteristic power signal, wherein evaluating the spectral correlation count includes counting a number of frequency components having a coherence value below a threshold.
[0019] In some embodiments, the coherence value may be below a threshold of 0.8. With a reasonable threshold (e.g., 0.8), any outliers in the spectral correlation counts that may be close to a 1 to 1 correlation can be excluded, because a 1 to 1 correlation indicates that there is no difference between the frequency components of the compared signals.
[0020] In some embodiments, the method may also include evaluating the phase difference between the aggregate power signal at 120 Hz and the load class signature power signal. Again, this is a distinguishing feature that helps provide easier and more accurate prediction of low power loads present in the aggregate signal.
[0021] In some embodiments, the method may further include calculating an overall probability using a sigmoid membership function for each feature, wherein the features include a VI waveform area difference, a spectral correlation count, and a phase difference between the aggregate power signal and the load category characteristic power signal.
[0022] In some embodiments, the sigmoid membership function may be given by the following equation:
[0023]
[0024] Where α is the center point of the distribution and β is the width of the distribution.
[0025] In some implementations, the overall coherence probability may be calculated using the following equation:
[0026] Prob=Phase120×DiffArea×SpectralCorr
[0027] Where Phase120 is the membership function of the phase difference at 120 Hz, DiffArea is the membership function of the area difference of the VI waveform, and SpectralCorr is the membership function of the spectral correlation count.
[0028] In some implementations, a low power mode may exist when the value of the coherence probability is greater than 0.5.
[0029] In some embodiments, the frequency components may include AC and DC. Using a single method to measure the frequency components of AC and DC covers all different types of loads.
[0030] In some implementations, the number of frequency components may be greater than>1.
[0031] In one aspect of the present disclosure, an operating mode detection system for overlapping electrical loads is provided, the operating mode detection system comprising: a power socket; an external computing component, wherein the external computing component is configured to: measure an aggregate power signal via the power socket; determine the operating mode of the aggregate power signal, wherein the operating mode is determined to be one of an active mode, a lower power mode, and an off mode; when it is determined that the operating mode is an active mode, the external computing component is further configured to: determine the load category of the aggregate power signal; select a corresponding load category characteristic power signal from a load category database; evaluate the spectral coherence between the aggregate power signal and the load category characteristic power signal; determine the overall coherence probability between the frequency components of the aggregate power signal and the frequency components of the load category characteristic power signal; based on the value of the determined coherence probability, determine whether a low power mode exists within the aggregate signal of the active mode; and determine the operating mode of each load in the overlapping loads. BRIEF DESCRIPTION OF THE DRAWINGS
[0032] The present disclosure will now be described, by way of example only, with reference to the accompanying drawings, in which:
[0033] Figure 1 is a schematic diagram of the system components;
[0034] Figure 2 is a flow chart of method steps for determining operating modes for overlapping and non-overlapping loads;
[0035] Figure 3 a graph showing actual power for multiple cycles of a load in active mode and a graph of mode ID;
[0036] 4a-c show graphs of actual power and mode ID for multiple cycles of 3 different loads in low power mode;
[0037] Figure 5 shows a graph of combined real power and a graph of combined mode ID for multiple cycles of 4 different loads in active mode and low power mode;
[0038] Figure 6 is a graph of the aggregate current of 4 different loads combined and the current of the individual loads, one load in active mode, and 3 loads in low power mode;
[0039] Figure 7a shows the voltage / current VI traces of four different loads, where all loads are in low power mode;
[0040] FIG7 b shows the aggregated VI traces of FIG7 a at four different loads;
[0041] Figure 8a shows the VI traces of four different loads, where one load is in active mode and the others are in low-power mode;
[0042] FIG8 b shows the aggregated VI traces of FIG8 a at four different loads;
[0043] Figure 9a shows the VI traces of four different loads, where one load is in low-power mode and the others are in active mode;
[0044] FIG9 b shows the aggregated VI traces of FIG9 a at four different loads;
[0045] Figure 10a shows the VI traces of four different loads, where all loads are in active mode;
[0046] Figure 10b shows the aggregated VI traces of Figure 10a at four different loads;
[0047] Fig.11 A graph showing example VI traces for different load categories;
[0048] FIG12 a shows a graph of the area distribution when a lower power is present in the polymerization current;
[0049] FIG12 b shows a graph of the area distribution when there is no lower power in the polymerization current;
[0050] FIG13 a shows a normalized graph comparing the sample signal to the aggregated current, power and VI traces, where all loads are in low power mode;
[0051] Figure 13b is the coherent spectrum of the current when all loads are in low power mode;
[0052] FIG14 a shows a normalized graph comparing a sample signal to an aggregated current, power, and VI trace signal, where one load is in active mode and the other loads are in low power mode;
[0053] FIG14 b is a coherent spectrum of the current when one load is in active mode and the other loads are in low power mode;
[0054] FIG15 a shows a normalized graph comparing a sample signal to an aggregated current, power, and VI trace, where one load is in a low power mode and the other loads are in an active mode;
[0055] FIG15 b is a coherent spectrum of the current when one load is in low power mode and the other loads are in active mode;
[0056] FIG16 a shows a normalized graph comparing the sample signal to the aggregated current, power and VI traces, where all loads are in active mode;
[0057] Figure 16b is the coherence spectrum of the current when all loads are in active mode;
[0058] FIG17a shows a graph of spectral correlation counts when relatively low power is present in the aggregate current;
[0059] FIG17 b shows a graph of spectral correlation counts when there is no lower power in the aggregate current;
[0060] FIG18a shows a graph of the phase difference at the 120 Hz frequency component when lower power is present in the aggregate current;
[0061] FIG18 b shows a graph of the phase difference at the 120 Hz frequency component when there is no lower power in the aggregate current;
[0062] Fig.19 A graph of the sigmoid function of the membership function used to determine the overall coherence probability is shown. DETAILED DESCRIPTION
[0063] Figure 1 is a schematic diagram showing components of the operating mode detection system 100. Figure 1As shown, there are multiple loads 102 connected to a single circuit breaker 104, that is, these loads have a shared circuit. Therefore, a single load does not have a dedicated circuit, that is, one load 112 is connected to one circuit breaker 110. It will be appreciated that multiple loads 102 can be connected to a single circuit breaker 104, and the operating mode detection method can be used on a system containing one or more loads 102. The circuit breaker 104 / 110 can be an intelligent circuit breaker, an energy management circuit breaker (EMCB), etc. The system can be extended to a large-scale building power management system, where there is a combination of dedicated circuits and non-dedicated circuits and connected to a load center 106. The system 100 also includes an acquisition and processing module 108, which is used to acquire and process electrical signals and power signals from the circuit breakers 104 / 110. The acquisition and processing module 108 can be any suitable computing device or multiple computing devices that can be operated to acquire electrical signals and process data according to the operating mode detection method.
[0064] Figure 2 2 is a flow chart 200 of method steps for determining operating modes of overlapping and non-overlapping loads. In particular, a method for real-time detection of loads in low power operating modes under overlapping load conditions is provided. Figure 2 As shown in the flowchart 200 of FIG. 1 , the method begins by reading the aggregate power signal (i.e., measuring the aggregate current signal and the aggregate voltage signal) at the power outlet or circuit breaker for a minimum of five cycles 202. Figure 1 The system's acquisition and processing module 108 is executed in real time. A minimum of 5 cycles are required to obtain an accurate reading.
[0065] Once the aggregate signal is acquired, the system evaluates the operating mode of the aggregate signal 204 by determining whether the operating mode is an active mode, a low power mode, or an off mode. The prior art method disclosed in U.S. Patent No. 10,837,989 is used to calculate the overall operating mode of the aggregate signal. If all loads connected to the socket are in low power mode, the aggregate mode is detected as low power mode. Therefore, vice versa, because if the aggregate mode is detected as low power mode, it is inferred that all connected loads are in low power mode. Similarly, if the aggregate current signal is zero, that is, no signal is read, all loads are off. When evaluating the operating mode, the method 200 determines whether the operating mode is an active mode 206 based on the aggregate data. If no active mode is found, the method 200 then determines whether there is a low power operating mode 208. Therefore, it is concluded that a low power mode exists on one or more loads, or all loads are off.
[0066] If it is determined that the operating mode of the aggregate signal is active mode 206, there may be a low power mode load connected along with the active power load. Therefore, further processing is required to classify the presence of a low power mode load within the aggregated active mode signal. First, the load category of the aggregate signal is determined 210. The load can be characterized as various load categories, including resistive (R), reactive (X), power electronics with power factor correction (P), power electronics without power factor correction (NP), microwave or complex load (M), transformer (T), phase angle controlled (PAC) load. The voltage / current (VI) trace waveform varies depending on the load category, so evaluating the VI trace of the aggregate signal will help identify the dominant load category of the aggregate signal. About Fig.11 The details of the VI traces for the various load categories are discussed further. Once the load category is determined 210, the voltage waveform and the aggregate current waveform of the aggregate signal are normalized 212. Normalizing the aggregate current and voltage waveforms between -1 and 1 enables evaluation and comparison of different loads at different power levels. Then, based on the identified load category, the normalized signal is compared to a sample waveform for that category, as in step 214 of the method flow chart 200. The sample waveform is selected from a library or database of current signatures for each of the different load categories 214. The sample waveform is a normalized waveform for one cycle in the active mode for the particular load category selected.
[0067] The following steps 216 to 228 are implemented to extrapolate a low power mode signal from the aggregate signal by evaluating various distinguishing features between the aggregate signal and the sample waveform from 214. In step 216, the distinguishing feature evaluated is the difference in the area enclosed by the VI traces of the aggregate signal and the sample signal. Note that the VI traces for a given load category of individual loads differ when compared to the aggregate load. Therefore, the root mean square error (RMSE) is used to determine the difference between the areas enclosed by the VI traces, which provides an indication of the presence of a low power mode within the aggregate signal. Further details on determining the area difference 216 are provided below with respect to Figures 12a and 12b.
[0068] The next distinguishing feature evaluated is the spectral coherence between the sample waveform and the aggregate waveform 218. In particular, the spectral coherence 218 is performed between the sample current of the sampled signal and the aggregate current of the aggregate signal, both of which have been pre-normalized. The spectral coherence 218 is determined using the Welch average modified periodogram method. The spectral coherence indicates the correlation between the various frequency components and their respective phase angles. When the load current includes all active mode loads, it is found that the spectral coherence is high for most frequency components, while when the load current includes a mixture of low power mode loads and active mode loads, the correlation is found to decrease. Therefore, the reduced correlation is an indicator of the presence of a load operating in a low power mode in the aggregate signal. Then, the coherence estimates of all frequency components of the signal are recorded 220, and the number of frequency components with a coherence value less than a threshold is calculated 222. The coherence value is preferably below a threshold of 0.8, and more preferably below a threshold of 0.6. Further details on the evaluation of spectral coherence are discussed with respect to Figures 13a to 17b.
[0069] Once the spectral coherence is determined, the next step of the method 200 is to determine the distinguishing features 224 of the phase difference between the aggregate signal and the sample signal. The phase difference is evaluated at the 120 Hz component, i.e., at the second harmonic. It is found that when one or more low power mode loads are present, the phase difference at 120 Hz is higher over a larger phase difference range (-200 to 200 degrees). Therefore, the increased distribution of the phase difference at 120 Hz is an indicator of the presence of low power operating mode loads in the aggregate signal. More details on the phase difference are provided below in conjunction with Figures 18a and 18b.
[0070] Once the difference in the distinguishing features is determined, as in steps 216 to 224, the method 200 then proceeds to determine an overall probability 226 that one or more loads are operating in a low power mode. The sigmoid membership function is used to determine whether a low power operating mode load is present in the aggregate signal by inputting data from the distinguishing features into the sigmoid function. If the overall probability is found to be greater than 0.5, then a low power mode is present in the aggregate signal 228. However, if the overall probability is less than 0.5, then a low power mode is not present and the aggregate signal includes all active mode loads. It will be appreciated that the probability threshold may be a value greater than 0.5, such as 0.6, 0.7, 0.8, etc., depending on the desired level of accuracy.
[0071] Combination Figure 2 The method 200 provides further details for each step in determining various factors required to determine the presence of a low power operating mode load in an aggregate signal.
[0072] Figure 3and FIGS. 4a - c show graphs 300 / 400 / 420 / 440 of actual power and mode ID for multiple cycles of various loads for different operating modes. Figure 3 300 by the actual power graph 302, the power characteristics over 10 cycles are characteristic of a sine wave (i.e., an AC signal), with the power being between approximately 120-135W. In addition, the mode classified as Mode 1 does not change over the entire cycle duration, as shown in FIG. Figure 3 4a-c show the behavior of different loads of various household appliances and devices when in standby mode or hibernation (i.e., operating in a low power mode). It can be clearly seen from the actual power graphs 402, 422 and 442 of Figures 4a-c that when compared with Figure 3 The power values have decreased when compared to 302 of FIG. 4a-c. The variation in power values on the graphs 402, 422, and 442 of FIG. 4a-c is due to the different currents drawn by different appliances / devices in standby mode. However, the graphs 402, 422, and 442 of FIG. 4a-c have similar characteristic power waveforms over a 10 cycle duration, where the power starts high and then decreases from cycle 1 to cycle 10. These waveforms are different from Figure 3 302. In addition, the mode classified as mode 3 does not change for each of the mode graphs 404 / 424 / 444 in FIGS. 4a-c.
[0073] Figure 5 The actual power graph 502 and the mode ID graph 504 are shown for an aggregate power signal including a mix of active operating modes and low power operating modes over 10 cycles. Figure 3 The active mode power characteristic 302 (i.e., characteristic AC sine wave) is comparable. In addition, the mode ID for the aggregate signal is classified as mode 1, i.e., active mode, as also shown in 504. Figure 3 304. Therefore, measuring the power over multiple cycles of the aggregated signal from multiple loads on the same circuit does not provide any insight into whether a low operating mode exists on any load, as the high power load dominates the overlapping loads. Therefore, this method cannot distinguish whether any load is in a low power mode, and therefore further analysis is required, such as Figure 2 The same is true for a load that transitions from active mode to low power mode in an overlapping load circuit where multiple loads are in active mode. Since the power change for the transition is very small, the transition may go unnoticed when the high power load that remains on dominates the aggregate signal. As a result, the switching of one load from active mode to low power mode is not detected. Therefore, the real-time execution of Figure 2 The method 200 is described to detect real-time load behavior.
[0074] Figure 6 600 is a graph of the aggregate current of 4 different loads in combination 606 and the current of each individual load 602 / 604 for 200 samples, where the loads are arranged so that one load is in active mode 602 and three loads are in low power mode 604. It can be clearly seen from the measured current signals of 600 that the aggregate current 606 or the currents 602 and 604 of the combined loads is comparable to the current of the load in active mode (AM) 602. Therefore, similar to Figure 3 and the actual power measurements in Figure 4a–c, the aggregate current pattern is dominated by the high power load.
[0075] As discussed with respect to step 210 of method 200, the load category is determined based on the VI trace of the aggregate signal. Figures 7a to 10b show the VI traces of 4 loads in various operating mode combinations to help resolve overlapping aggregate signals. Figure 7a shows the individual VI traces 700 of all 4 loads 702 / 704 / 706 / 708 when all loads are operating in low power mode. Figure 7b is the combined VI trace 720 of the aggregate signal of the loads from Figure 7a. As expected for all loads in low power mode, the VI trace 720 of the aggregate signal is comparable to the VI traces 700 of the individual loads 702 / 704 / 706 / 708. Figure 8a shows each load VI trace in the individual load VI traces 800 when one load is in active mode 802 and the remaining loads are in low power mode 804 / 806 / 808. FIG8b is a combined VI trace 820 of the aggregate signal measured when the loads are operating as in 800 of FIG8a, i.e., one load is in active mode 802 and the other loads are in low power mode 804 / 806 / 808. Clearly, the aggregate VI trace 820 is dominated by the VI trace 802 of the active mode load, as the shapes of the VI traces are comparable. FIG9a and FIG9b represent VI traces 900 / 920 when one load is in low power mode 902 and the remaining loads are in active mode 904 / 906 / 908. Likewise, the active mode loads 904 / 906 / 908 of FIG9a dominate the aggregate VI trace 920 of FIG9b. FIG10a and FIG10b are individual VI traces 1000 and combined VI trace 1020, respectively, for the case where all loads 1002 / 1004 / 1006 / 1008 are operating in active mode. The combined VI trace 1020 in FIG10b has a shape that is a combination of the load VI traces 1000 of FIG10a.
[0076] Further analysis of various loads with different load categories shows that the resulting load category of the aggregate load represents the dominant load in the circuit. Therefore, the VI trace corresponding to each load category can be used to determine the load category of the aggregate signal, such as in step 210 of method 200, and for analysis in steps 214 and 216. The VI traces of different load categories are shown in FIG. Fig.11 1100. VI trajectories for the following load categories are shown: power electronic load 1102 without power factor correction (NP); power electronic load 1104 with power factor correction (P); resistive (R) 1106; reactive (X) 1108; complex (M) 1110; phase angle controlled (PAC) 1112 / 1114; transformer (T) 1116; and low power mode (LP) 1118. Two PAC loads 1112 / 1114 are presented for loads with different operating phase angles. PAC load 1112 is controlled at a phase angle >90 degrees, and PAC load 1114 is controlled at a phase angle <90 degrees. Depending on the operating phase angle of the PAC device when in use, the VI trajectories may be different. Fig.11 Different VI traces for PAC load 1112 and PAC load 1114 are shown. Figure 2 As discussed in method 200, based on Fig.11 1100, a load category is selected from the evaluated VI trace 210, the voltage and aggregate current waveforms are normalized 212, and a sample signal of the corresponding load category is selected 214. For example, if the aggregate signal is found to be closest to load category X, as in 1108, then the sample signal of the load of category X is used for comparison. Before the analysis begins, both the aggregate signal and the sample signal are normalized in order to compare the characteristics of loads of the same category but operating in different power ranges.
[0077] According to step 216 of method 200, the area difference between the VI trace of the aggregate signal and the VI trace of the sample signal should result in different area distributions, as shown in graphs 1200 and 1220 of Figures 12a and 12b. The aggregate signal includes signals corresponding to at least 2 or more loads, so when the aggregate signal is compared with the individual sample signals, there will be differences in the VI traces. In addition, these differences in the VI traces and area distributions will be different for the aggregate signal in the presence of a low power mode and the aggregate signal in the absence of a low power mode, as shown in 1200 in Figure 12a and 1220 in Figure 12b, respectively. The total area distribution when the low power mode 1200 is present is higher than the total area distribution when the low power mode 1220 is not present, as shown in Figures 12a and 12b. Therefore, when the appliance is in low power mode, when compared to the active mode, evaluating the area enclosed within the VI trace (i.e., graphs 1200 and 1220) provides a clear distinction in area values. By evaluating the area difference of the VI traces of the aggregate signal and the sample signal, it can be determined whether a low power operating mode exists in the aggregate signal. As described above, the area difference is a distinguishing feature used to determine the overall probability of the loads operating in the low power mode based on the signals of the overlapping loads.
[0078] Another distinguishing feature analyzed is spectral coherence, as in steps 218-222 of method 200. Spectral coherence indicates the correlation between various frequency components and their respective phase angles. Method 200 calculates spectral coherence for multiple frequency components including DC (0 Hz). For spectral coherence analysis, the number of frequency components is greater than 1. In the example below, the number of frequency components is 129.
[0079] 13a-16b illustrate normalized graphs 1300 / 1400 / 1500 / 1600 comparing sample and aggregate signals of current (e.g., 1302 and 1308), power (e.g., 1304 and 1310), and VI traces (e.g., 1306 and 1312), and corresponding coherent spectra of current 1320 / 1420 / 1520 / 1620 when the load is in a particular operating mode configuration. Graphs 1300 and 1320 of FIGS. 13a and 13b are for the case when all loads are in low power mode; graphs 1400 and 1420 of FIGS. 14a and 14b are for the case when one load is in active mode and the other loads are in low power mode; graphs 1500 and 1520 of FIGS. 15a and 15b are for the case when one load is in low power mode and the other loads are in active mode; and graphs 1600 and 1620 of FIGS. 16a and 16b are for the case when all loads are in active mode. It can be inferred from FIGS. 13a to 16b that when the load current of the aggregate signal includes all loads operating in active mode, the spectral correlation is higher for most frequency components. For the case where overlapping loads operate in a combination of both low power mode and active mode, there are some frequency components where the correlation is reduced. Therefore, for loads operating in active mode, lower spectral correlation values are expected as more frequency components are closely correlated. Therefore, analyzing the spectral coherence of the aggregate signal and the sample signal provides a further distinguishing feature for detecting the presence of low power modes in the aggregate signal from overlapping loads.
[0080] Similar to the analysis of the area difference of the VI traces in 1200 and 1220 of Figures 12a and 12b, the distribution of the correlation components of the aggregate signal with and without low power is also analyzed. The frequency components with coherence values below a threshold value of 0.8 are counted, i.e., the number of correlated frequency components with coherence values <0.8. Figure 17a shows a spectrum correlation count chart 1700 when there is lower power in the aggregate current of the overlapping load (i.e., low operating mode). Figure 17b shows a spectrum correlation count chart 1720 when there is no lower power in the aggregate current. In contrast, the chart 1700 of Figure 17a shows that the total number of frequency components with coherence values below the threshold has a higher value than the total number of the chart 1720 of Figure 17b. In particular, when comparing the right sides of the count charts 1700 and 1720, the chart 1700 has a high probability value on a higher number (i.e., >30) of correlation components, while the chart 1720 has a smaller probability for the number of correlation component values >30. Therefore, for a load operating in low power mode in the aggregate signal, the overall count distribution of frequency components having coherence values less than the threshold value (ie, 0.8) is higher than when the load operates only in active mode.
[0081] Another distinguishing feature used to determine whether one or more loads are operating in a low power mode based on the aggregate signal is the phase difference at 120 Hz, as shown in Figure 2 FIG. 18a shows a count distribution 1800 of phase differences at 120 Hz for a selected number of frequency components when lower power is present in the signal of the aggregated current. FIG. 18b shows a count distribution 1820 of phase differences at 120 Hz for a selected number of frequency components when lower power is not present in the signal of the aggregated current. Obviously, when a low power mode is present in the signal of the aggregated current, the number of count distributions with a larger phase difference range is larger, that is, the number of counts at the upper end (200 degrees) and the lower end (-200 degrees) of the phase difference is higher in graph 1800 than in graph 1820. Therefore, a load in a low power mode will produce an overall higher phase difference range value than when the load is in an active mode. Therefore, evaluating the phase difference at 120 Hz helps determine whether there are one or more low power mode loads in the overlapping aggregated signal.
[0082] Using all the above distinguishing features and their distribution plots (i.e., area difference, spectral correlation, and phase difference at 120 Hz), an equation based on the sigmoid membership function is used to detect the overall probability of the presence of a low power operating mode within the acquired overlapping aggregate signal. The sigmoid membership function is given by the following equation:
[0083]
[0084] Where α is the center point of the distribution and β is the width of the distribution. The sigmoid function 1900 defined by the above equation is Fig.19 , and for each distinguishing feature. The values of α and β are derived during training, and as new readings are collected in real time, the function for each corresponding feature is calculated and the x value is determined. The resulting coherence probability is calculated as the multiplication of the output of each membership function, as calculated using the following equation:
[0085] Prob=Phase120×DiffArea×SpectralCorr
[0086] Where Phase120 is the membership function of the phase difference at 120 Hz, DiffArea is the membership function of the area difference of the VI waveform, and SpectralCorr is the membership function of the spectral correlation count.
[0087] If the value of the coherence probability is greater than 0.5, it is concluded that no load (among the multiple loads) in the aggregated current is running in low power mode. If the value of the coherence probability is less than 0.5, it indicates that one or more loads (among the multiple loads) in the aggregated current are running in low power mode. Therefore, the results of the analysis and determination of whether there is a low power mode in the signals of overlapping loads can be used to provide power management suggestions to the user. For example, the aforementioned method and system can highlight the circuits where there are loads running in low power mode, and therefore if not actively used, they can be turned off.
[0088] Those skilled in the art will recognize that other distinguishing features (e.g., features from the time domain and / or frequency domain) can be evaluated to increase accuracy in determining the operating mode of overlapping electrical loads. In addition, the example analysis throughout this disclosure is based on 1 to 4 load changes when determining distinguishing features. However, those skilled in the art will recognize that an increase in the number of loads on a single circuit and an increase in the number of circuits will further enhance the results of this method.
Claims
1. A method for detecting an operation mode of an overlapping electric load, the method comprising: Measure the aggregate power signal of the power outlet; determining an operating mode of the aggregate power signal, wherein the operating mode is determined to be one of an active mode, a lower power mode, and an off mode; When it is determined that the operating mode is the active mode, the method further includes: determining a load category of the aggregate power signal; Selecting a corresponding load category characteristic power signal from a load category database; evaluating spectral coherence between the aggregate power signal and the load category characteristic power signal; determining an overall coherence probability between frequency components of the aggregate power signal and frequency components of the load category characteristic power signal; determining whether a low power mode is present within the aggregate power signal of the active mode, whereby the low power mode is arranged to be detectable based on the value of the coherence probability; and An operating mode of each of the overlapping electrical loads is determined.
2. The method according to claim 1, wherein: Measuring the aggregate power signal includes reading an aggregate current signal and an aggregate voltage signal of the power socket.
3. The method according to claim 2, wherein: Reading the aggregate current signal and the aggregate voltage signal is performed for a minimum of 5 cycles.
4. The method according to claim 1, wherein: The load category of the aggregate power signal is determined to be one of a plurality of load categories, the load categories comprising: Power electronic loads without power factor correction (NP); Power electronic loads with power factor correction (P); Transformer (T); reactive (X); Phase Angle Controlled (PAC); complex (M); or Resistance (R).
5. The method according to claim 1, further comprising normalizing the aggregate power signal, wherein: The aggregated current and voltage waveforms were normalized.
6. The method according to claim 1, wherein: The load category characteristic power signal includes a normalized waveform in an active mode for one cycle of each load category.
7. The method according to claim 1, wherein: Evaluating the spectral coherence between the aggregate power signal and the load category characteristic power signal further includes: The difference between the aggregate power signal and the load category characteristic power signal is evaluated based on their respective voltage / current, VI trace waveform, and area enclosed in the VI trace.
8. The method according to claim 7, wherein: Evaluating the spectral coherence between the aggregate power signal and the load category characteristic power signal further includes: Using root mean square error (RMSE) analysis, the difference between the aggregate power signal and the area enclosed by the VI trace of the load category signature power signal is calculated.
9. The method according to claim 8, wherein: Evaluating the spectral coherence between the aggregate power signal and the load category characteristic power signal further includes: The spectral coherence between the normalized current signal of the aggregate power signal and the normalized current signal of the load category characteristic power signal is estimated using a Welch average modified periodogram method.
10. The method according to claim 9, wherein: Evaluating the spectral coherence between the aggregate power signal and the load category characteristic power signal further includes: An estimate of the spectral coherence of all frequency components of the compared signals is recorded.
11. The method according to any one of claims 1 to 10, further comprising evaluating a spectral correlation count between the aggregate power signal and the load category characteristic power signal, wherein: Evaluating the spectral correlation count includes counting the number of frequency components having a coherence value below a threshold.
12. The method according to claim 11, wherein: The coherence value is below a threshold value of 0.
8.
13. The method according to any one of claims 1 to 12, further comprising: The phase difference between the aggregate power signal and the load category characteristic power signal at a frequency of 120 Hz is evaluated.
14. The method according to any one of claims 1 to 13, further comprising calculating the overall probability using a sigmoid membership function for each feature, wherein, The characteristics include a VI waveform area difference, a spectrum correlation count, and a phase difference between the aggregate power signal and the load category characteristic power signal.
15. The method according to claim 14, wherein: The sigmoid membership function is given by the following equation: Where α is the center point of the distribution and β is the width of the distribution.
16. The method according to any one of claims 1 to 15, wherein: The overall coherence probability is calculated using the following equation: Prob=Phase120×DiffArea×SpectralCorr Where Phase120 is the membership function of the phase difference at 120 Hz, DiffArea is the membership function of the area difference of the VI waveform, and SpectralCorr is the membership function of the spectral correlation count.
17. The method according to any one of claims 1 to 16, wherein: When the value of the coherence probability is greater than 0.5, a low power mode exists.
18. The method according to any one of claims 1 to 17, wherein: The frequency components include AC and DC.
19. The method according to any one of claims 1 to 18, wherein: The number of frequency components is greater than>1.
20. An operation mode detection system for overlapping electrical loads, the operation mode detection system comprising: Power socket; An external computing component, wherein the external computing component is configured to: measuring an aggregate power signal via the power outlet; determining an operating mode of the aggregate power signal, wherein the operating mode is determined to be one of an active mode, a lower power mode, and an off mode; When the operating mode is determined to be an active mode, the external computing component is further configured to: determining a load category of the aggregate power signal; Selecting a corresponding load category characteristic power signal from a load category database; evaluating spectral coherence between the aggregate power signal and the load category characteristic power signal; determining an overall coherence probability between frequency components of the aggregate power signal and frequency components of the load category characteristic power signal; determining whether a low power mode exists within the aggregate signal of the active mode based on the determined value of the coherence probability; and Determine the operating mode for each of the overlapping loads.
