A non-intrusive electrical appliance identification method, system, medium, device and terminal

Abstract

The present invention belongs to the technical field of household appliance identification, and discloses a non-invasive electrical appliance identification method, system, medium, equipment and terminal. The current of the electrical appliance is synchronously sampled by using a current transformer and a voltage zero-crossing detection circuit; the synchronously sampled current data is processed by using an MCU to obtain the parameters of the fundamental phase of the current, the amplitude ratio of each harmonic to the fundamental wave, and the energy ratio, and the characteristic quantity of each electrical appliance is calculated by using the current parameters; the type of the electrical appliance that has changed is determined by extracting dynamic characteristic values ​​from the changing data, and the type and quantity of the current electrical appliance are intelligently identified by combining accumulation and CUSUM. The present invention takes into account low cost, low power consumption, and high recognition accuracy, and is easy to industrialize and promote; it has a learning function, can autonomously learn new electrical appliances, and the identification of electrical appliances is not limited to the types of electrical appliances built-in at the time of leaving the factory; training a small amount of data for a single electrical appliance can accurately identify the combined electrical appliances.

Description

Technical Field

[0001] The present invention belongs to the technical field of household appliance identification, and in particular relates to a non-intrusive household appliance identification method, system, medium, device and terminal. Background Art

[0002] In recent years, with the rapid development of social economy, industrial power energy consumption and household power energy use have increased sharply. Facing the power pressure in the new era, it is imperative to upgrade and optimize the power system. In the field of residential electricity consumption, the Internet of Things power control system is specifically implemented as household appliance status monitoring, power consumption notification, and intelligent control of electrical appliances.

[0003] In the face of these household power monitoring needs, basic requirements such as identifying the type of electrical appliances, detecting the use status of electrical appliances, and monitoring power usage are constantly being valued by people. In terms of electrical appliance identification and household load monitoring, there are two main research methods, namely invasive and non-invasive. Compared with invasive load monitoring technology, non-invasive power load monitoring technology does not require cumbersome hardware installation. It only needs to collect the voltage and current at the power supply entrance, and the category status and power consumption of electrical appliances can be obtained through specific algorithms. Considering the high cost of invasive monitoring, non-invasive monitoring has become the mainstream direction of household power monitoring. Non-invasive monitoring aims to identify the use status of electrical appliances by processing household aggregated power consumption signals. However, in the non-invasive electrical appliance monitoring and identification method, relying solely on steady-state or transient parameters cannot accurately determine the type of electrical appliances, and relying solely on high-precision harmonic analysis methods will produce data redundancy.

[0004] Cheng Chunyu et al. proposed a design of an electrical appliance identification system based on BP neural network. The system collects the working current data of a single electrical appliance and transmits it to the host computer for storage and processing. The host computer performs composite digital filtering on the received data, and arranges and combines multiple situations based on the BP neural network through the superposition principle to generate a training data set, thereby realizing electrical appliance identification.

[0005] Cao Yilong et al. proposed an electrical appliance identification method based on FFT and genetic algorithm. They analyzed the current waveform spectrum of typical electrical appliance loads and used its harmonics as the optimal characteristic parameters for load identification. The waveform data was processed through fast Fourier transform, and the minimum Euclidean distance was used as the optimization indicator. A genetic algorithm was used to search for the optimal solution, ultimately achieving accurate identification of the electrical appliance category.

[0006] Zhou Xiao et al. proposed a non-intrusive power load identification algorithm based on ELM. Starting from the load characteristics, the transient and steady-state electrical characteristics of each power load are extracted and established. Then, the extreme learning machine (ELM) neural network model is used to nonlinearly map the input features to the input layer to achieve rapid convergence to the global optimal point. The bilateral event detection method based on cumulative sum (CUSUM) is used to quickly and accurately detect load switching events and trigger load identification in real time.

[0007] The above solutions all use neural network or artificial intelligence algorithms to achieve high-precision identification of electrical appliances. At the same time, they all require complex training and complex calculations, which leads to increased system power consumption and increased costs.

[0008] Although the electrical appliance recognition algorithms proposed by Cheng Chunyu, Cao Yilong, Zhou Xiao and others use different specific recognition methods, they all use neural network or artificial intelligence algorithms to achieve high-precision requirements for electrical appliance recognition. Although the cost is controlled to a certain extent, due to the inherent training and computing requirements of neural network or artificial intelligence algorithms, the increase in system power consumption and cost is inevitable. It is difficult to meet the requirements of high precision while taking into account the requirements of low cost. The cost issue is particularly important in the field of household appliances, which will undoubtedly limit the actual promotion and use of these devices and cannot perfectly meet the needs of market users.

[0009] In addition to these systems that use neural networks or artificial intelligence algorithms, other lower-cost recognition devices have the obvious problem of insufficient recognition accuracy. They can only recognize some electrical appliances fixed by manufacturers, with a very limited scope of use, or the recognition accuracy is not high, and it is impossible to achieve high-accuracy real-time recognition display.

[0010] Prior art 1 (patent number 201710448867.7) discloses a non-intrusive residential load identification method based on S_Kohonen, comprising the following steps: Step 1: judging a switching event based on changes in active power at a household power inlet, and when a switching event occurs, collecting electrical current samples of the switching event at the household power inlet; Step 2: performing frequency domain analysis on the collected electrical current samples, extracting the frequency domain harmonic amplitude as the load characteristics of each electrical appliance, and forming a load characteristic library; Step 3: designing a S_Kohonen neural network suitable for resident load identification, determining the number of neurons in the input layer and output layer of the S_Kohonen neural network and the size of the competition layer, so as to determine the selection method of the winning neuron and the learning algorithm for weight adjustment; Step 4: initializing parameters; Step 5: training the S_Kohonen network with a training set and testing it with a test set; Step 6: adjusting network parameters to achieve optimal network performance. However, this technology is only effective for electrical appliances above 20W. It cannot correctly judge the switching events of appliances below 20W, and thus cannot identify such appliances. It only uses power changes as the basis for judging switching events and uses artificial intelligence algorithms, which requires more calculations and consumes more system power.

[0011] Prior art 2 (patent number 201810133757.6) discloses a non-invasive electrical appliance identification method based on the Boost model, including the following steps: 1. Use unlabeled data collection to collect meter data; 2. Guide the user to perform switch operations on the electrical appliances to be identified, and record the meter power data and the corresponding electrical appliance labels; 3. First, use the Bootstrap algorithm to expand the labeled set, combine it with the unlabeled set into a training set, and then use it to establish a classification model; 4. Collect power mutation events in real time, classify them with the trained model, and display the identified electrical appliance results to the user; 5. Add the new samples and new labels in the pre-identification stage to the original labeled data set, and repeat steps 3 and 4; 6. Use a model that meets the accuracy requirements to identify electrical appliances, and provide real-time feedback to users or demanders. However, this technology uses power changes as the basis for judging switching events, and its applicable electrical powers are all above tens of watts, that is, it is only effective for high-power electrical appliances.

[0012] Prior art three (patent number 202111501733.X) discloses a load identification method based on a non-invasive intelligent terminal, which collects original load data through a collection terminal; uses a load event detection algorithm to locate the load switching point, and extracts stable segment data based on the switching point; performs data preprocessing to eliminate undesirable data; extracts features and constructs a load steady-state feature set; calculates feature weights of different features based on the entropy weight method; extracts the steady-state features of the unknown load, and calculates the improved feature weighted Euclidean distance between the unknown load and the known steady-state feature set based on the feature weights; takes the inverse of each type of sample as the voting weight, calculates the similarity between the unknown load and the feature set based on the voting weight, and takes the most similar load type as the discrimination result.

[0013] In today's era of massive use of electric energy, power monitoring is an important topic. In the field of household electricity, the requirements for household appliance identification are constantly being paid attention to. Existing appliance identification algorithms and systems often cannot take into account the two important requirements of low cost and high accuracy, and are not suitable for actual household appliance identification usage. The existing non-invasive appliance identification system has too high computational complexity, which is not conducive to low power consumption and industrialization.

[0014] As household identification devices, most of the devices currently available on the market are relatively low-cost non-invasive electrical appliance monitoring devices. In order to save costs while ensuring the accuracy of identification, today's monitoring devices mostly use some neural network or artificial intelligence algorithms, which require complex training and complex calculations, resulting in increased system power consumption and increased costs. Of course, there are also some devices that use simple identification methods, but it is difficult to maintain a good recognition effect, and the actual use effect is worrying. In general, in the field of household appliance identification, it is difficult to achieve both high precision and low cost. Therefore, it is urgent to design a new, low-complexity, high-precision method and system for non-invasive electrical appliance monitoring and identification.

[0015] Through the above analysis, the problems and defects of the prior art are as follows:

[0016] (1) Existing electrical appliance identification algorithms and systems often fail to take into account the two important requirements of low cost and high accuracy, and are not suitable for actual household appliance identification usage; the existing non-invasive electrical appliance identification system has too high computational complexity, which is not conducive to low power consumption and industrialization.

[0017] (2) The existing electrical appliance identification cost is high, and other lower-cost identification devices have the obvious problem of insufficient identification accuracy; or they can only identify electrical appliances fixed by the manufacturer, with a very limited scope of use, or the identification accuracy is not high, and it is impossible to achieve high-accuracy real-time identification display.

[0018] (3) Existing monitoring devices mostly use neural network or artificial intelligence algorithms, which require complex training and calculations, resulting in increased system power consumption and increased costs. Of course, there are also devices that use simple recognition methods, but it is difficult to maintain a good recognition effect, and the actual use effect is worrying.

[0019] (4) The existing non-intrusive residential load identification method is only effective for electrical appliances above 20W, and therefore cannot identify similar electrical appliances; it only uses power changes as the basis for judging switching events, and the use of artificial intelligence algorithms requires more calculations and the system consumes more power.

[0020] (5) The existing non-intrusive electrical appliance identification method uses power change as the basis for judging the switching event. The applicable electrical power is more than tens of watts, that is, it is only effective for high-power electrical appliances. Summary of the invention

[0021] In view of the problems existing in the prior art, the present invention provides a non-invasive electrical appliance identification method, system, medium, device and terminal, and more particularly, relates to a non-invasive electrical appliance identification method, system, medium, device and terminal based on dynamic characteristic parameters.

[0022] The present invention is implemented as follows: a non-invasive electrical appliance identification method, the non-invasive electrical appliance identification method comprising: using a current transformer and a voltage zero-crossing detection circuit to synchronously sample the current of the electrical appliance; using an MCU to process the synchronously sampled current data of 500ms as a group to obtain the parameters of the fundamental phase of the current, the amplitude ratio of each harmonic to the fundamental wave, and the energy ratio, and using the current parameters to calculate the characteristic quantity of each electrical appliance; by extracting dynamic characteristic values ​​from the changing data, the type of the changing electrical appliance is determined, and the type and quantity of the current electrical appliance are intelligently identified in combination with accumulation and CUSUM.

[0023] Furthermore, the non-intrusive electrical appliance identification method comprises the following steps:

[0024] Step 1: Initialize ADC and MCU to provide the necessary environment for data collection;

[0025] Step 2: Determine the current mode: determine whether to enter the learning mode; if entering the learning mode, enter the appliance number and execute step 3; if not entering the learning mode, enter the detection mode and execute step 5; this step completes the determination of the system working mode;

[0026] Step 3, determine whether the learning button is pressed: if the learning button is pressed, data is collected synchronously and characteristic parameters are calculated and stored, prompting that the learning is completed and displaying the characteristic parameters; if the learning button is not pressed, execute step 4; this step is used to learn the characteristics of the new electrical appliance to create conditions for intelligent identification of the device;

[0027] Step 4: Determine whether to exit the learning mode: If exiting the learning mode, enter the detection mode and execute step 5; if not exiting the learning mode, return to step 3 and re-perform the learning operation; this step is used to continuously learn the characteristics of new electrical appliances or exit the learning mode;

[0028] Step 5, start the detection operation: collect data regularly, calculate the characteristic parameter Vt, and determine whether the change of Vt exceeds the threshold; if not, it means that the number and type of the currently identified electrical appliances have not changed, the identification result remains unchanged, and return to step 2; if yes, it means that the number and type of the currently identified electrical appliances have changed, and the identification result needs to be updated, and execute step 6; this step is used to determine the change of the state of the electrical appliances, so as to trigger the identification process in time;

[0029] Step six, calculate the dynamic feature parameters, use the Euclidean distance to identify the changed electrical appliances, and update the type and quantity of the current electrical appliances based on the cumulative sum CUSUM; this step completes the identification of electrical appliances.

[0030] Furthermore, the non-intrusive electrical appliance identification method further includes:

[0031] When the system starts, the OLED screen will display the system function menu. The matrix keyboard will be used to select the system to enter the learning mode or detection mode. In the learning mode, the number of electrical appliances is 1. After the electrical appliances are enabled, the current waveforms and calculated characteristic parameters of each electrical appliance are collected and learned in turn through human-computer interaction, and these characteristic parameters are stored. In the monitoring mode, the system will periodically collect power consumption data and calculate the characteristic parameter V of power consumption. t , and according to V t The change in the status of electrical appliances is used to determine whether there is a change in the status of electrical appliances. If so, the dynamic feature parameters are calculated using the changed data and the Euclidean distance is used to determine the type of electrical appliances that have changed; the current type and quantity of electrical appliances are identified based on accumulation and update and displayed on the OLED screen.

[0032] Furthermore, the technical process of processing the parameters of electrical appliances in the non-invasive electrical appliance identification method is divided into acquiring the parameters of electrical appliances and extracting the change status of electrical appliances.

[0033] In the process of acquiring the parameters of electrical appliances, the MCU reads the current and voltage information synchronously collected by ADS1256, and then analyzes and obtains useful data; the MCU reads the data of the ADC module through the serial peripheral interface SPI; the sampling rate is set to 7.5K, and the MCU interrupt is triggered by the DRDY signal of ADS1256 to achieve continuous acquisition of the signal; each time the rising edge of the zero-crossing signal triggers the ADC to collect a group of data, and a circular FIFO queue is used to store the electrical parameter information.

[0034] In the process of extracting the state of electrical appliances, the current spectrum characteristics are extracted through FFT operation. Suppose the current sampling data of the tth group is f t (l), then the result of DFT at N points is:

[0035]

[0036] Furthermore, the non-intrusive electrical appliance identification method further includes:

[0037] When multiple electrical appliances are working at the same time, the 50Hz fundamental frequency and harmonics in the collected current spectrum are defined as:

[0038]

[0039] Where fs is the sampling rate, Denotes floor rounding operation. Define Q t As an evaluation function:

[0040]

[0041] The change value ΔQ of the data of the interval m group is used t As a preliminary basis for judging whether electrical appliances have changed:

[0042]

[0043] Wherein, H is the threshold, H=50mA, m=2; the interval m is adjusted according to the actual situation.

[0044] The difference method is used, and the amplitude difference ΔD of the spectrum at 50Hz is selected. t (1) As additional judgment conditions:

[0045] ΔD t (1)=|D t (1)|-|D t-k (1)|;

[0046] By comparing ΔD t (1) The value distinguishes the increase or decrease of high-power electrical appliances.

[0047] Select the power factor angle φ t (k) As one of the characteristic parameters for identifying changes in electrical appliances:

[0048] φ t (k) = arg(D t (k));

[0049] Here, arg(x) means the angle of x.

[0050] The function used to judge the change of electrical appliance status is:

[0051]

[0052] Furthermore, after processing the electrical appliance parameters, the electrical appliance identification method is as follows:

[0053] (1) Define the ratio of the amplitude of each harmonic of the electrical current to the amplitude of the 50 Hz fundamental frequency as

[0054]

[0055] make Eigenvector representing the harmonic ratios.

[0056] (2) Define the eigenvector Z t :

[0057]

[0058] in, is the effective value of current:

[0059]

[0060] (3) In the learning mode, the characteristics of a single electrical appliance are identified multiple times to obtain multiple feature vectors Z1, Z2, Z3, etc., which are used as the feature cluster C of the electrical appliance. i :

[0061]

[0062] Among them, N c is the number of sample electrical appliances, N i is the number of learning times of the i-th electrical appliance.

[0063] (4) In monitoring mode, the feature vector d of a set of unknown samples whose current electrical state remains unchanged is obtained:

[0064]

[0065] When a sample d is given, the minimum square error is obtained by dividing the characteristic clusters of electrical appliances into C = {C1, C2, C3, ...}:

[0066]

[0067] Among them, u i It is cluster C i The mean vector of :

[0068]

[0069] E describes the similarity between the sample and the cluster mean vector. The smaller the E value, the higher the similarity between the sample and the corresponding electrical appliance. By determining the value of the corresponding electrical appliance number i when the E value is the smallest, the type of the changed electrical appliance can be determined.

[0070] (5) After identifying the type of electrical appliances that have changed, according to ΔQ t Determine the state of the electrical appliance; when ΔQ t >0, the electrical appliance is turned on, when ΔQ t <0, the electrical appliance is turned off; based on the accumulation and principle, the type and quantity of the current electrical appliance are updated, and the OLED display output is updated.

[0071] Another object of the present invention is to provide a non-intrusive electrical appliance identification system using the non-intrusive electrical appliance identification method, wherein the non-intrusive electrical appliance identification system includes an MCU;

[0072] The MCU adopts STM32H7A3, and the peripherals are composed of a current sampling module, a voltage zero-crossing detection module, a power module, an ADC sampling module, an OLED display module and a keyboard module.

[0073] The current sampler is used to sample the current of electrical appliances and the zero-crossing detection circuit is used to detect the moment when the mains voltage crosses zero; the ADC with built-in PGA is used to amplify and convert the signal into analog-to-digital, and the MCU is used to synchronously collect current signals and intelligently detect the types and quantities of electrical appliances at the moment when the voltage crosses zero, and the results are displayed on the OLED screen.

[0074] Another object of the present invention is to provide a computer device, the computer device comprising a memory and a processor, the memory storing a computer program, and when the computer program is executed by the processor, the processor performs the following steps:

[0075] The current of electrical appliances is synchronously sampled using a current transformer and a voltage zero-crossing detection circuit. The synchronously sampled current data of 500ms is processed by an MCU to obtain the parameters of the fundamental phase of the current, the amplitude ratio of each harmonic to the fundamental, and the energy ratio, and the characteristic quantity of each electrical appliance is calculated using the current parameters. The type of electrical appliance that has changed is determined by extracting dynamic characteristic values ​​from the changing data, and the type and quantity of the current electrical appliances are intelligently identified by combining accumulation and CUSUM.

[0076] Another object of the present invention is to provide a computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, the processor executes the following steps:

[0077] The current of electrical appliances is synchronously sampled using a current transformer and a voltage zero-crossing detection circuit. The synchronously sampled current data of 500ms is processed by an MCU to obtain the parameters of the fundamental phase of the current, the amplitude ratio of each harmonic to the fundamental, and the energy ratio, and the characteristic quantity of each electrical appliance is calculated using the current parameters. The type of electrical appliance that has changed is determined by extracting dynamic characteristic values ​​from the changing data, and the type and quantity of the current electrical appliances are intelligently identified by combining accumulation and CUSUM.

[0078] Another object of the present invention is to provide an information data processing terminal, which is used to implement the non-intrusive electrical appliance identification system.

[0079] In combination with the above technical solutions and the technical problems solved, please analyze the advantages and positive effects of the technical solutions to be protected by the present invention from the following aspects:

[0080] First, in view of the technical problems existing in the above-mentioned prior art and the difficulty of solving the problems, the technical solutions to be protected by the present invention and the results and data during the research and development process are closely combined to analyze in detail and deeply how the technical solutions of the present invention solve the technical problems, and some creative technical effects brought about after solving the problems. The specific description is as follows:

[0081] The present invention designs a low-complexity, high-precision algorithm and system for non-invasive electrical appliance monitoring and identification. The system uses a current transformer and a voltage zero-crossing detection circuit to synchronously sample the current of electrical appliances, extract dynamic characteristic values ​​from the changing data, determine the type of electrical appliances that have changed, and finally combine the cumulative sum (CUSUM) to intelligently identify the type and quantity of current electrical appliances. At the same time, the present invention can accurately identify multiple household appliances that are used at the same time, and monitor the household electricity usage status in real time. At the same time, it provides certain conditions for realizing ubiquitous power Internet of Things client-side intelligent perception, and can be widely used in low-voltage power grids to realize diversified functions such as household appliance status monitoring and electricity usage notification.

[0082] The present invention synchronously collects current consumption on the power bus (at the household electricity meter) without affecting the working state of the electrical appliance itself; determines the state change of the electrical appliance through the self-defined multi-parameter detection method of the present invention, and decides whether to implement a specific identification process; after confirming that the type of the electrical appliance has changed, the present invention extracts dynamic feature parameters based on the changed information, and then uses the Euclidean distance to determine the type of the changed electrical appliance, and finally combines the cumulative sum (CUSUM) to realize the type and quantity of the current electrical appliance; while achieving low power consumption, a high recognition rate is guaranteed, and the present invention has an autonomous learning function, and can identify a variety of mixed combinations of electrical appliances (including a mixture of large and small power appliances), with a final recognition accuracy rate of up to 98.8%.

[0083] The prior art is only effective for electrical appliances above 20W, and cannot correctly judge the switching events of electrical appliances below 20W, thereby failing to identify such electrical appliances; Figure 5 It can be seen from the simulation results and Table 1 that the present invention can realize the identification of small electrical appliances of about 1W by adopting the function Vt of electrical appliance state change; the prior art only adopts power change as the basis for judging the switching event; the present invention adopts the parameter Vt that combines multiple change combinations such as power change, spectrum change, power factor change, etc. as the basis for judging the switching event, and has higher sensitivity; the prior art adopts artificial intelligence algorithm, which has a larger amount of calculation and system power consumption; the present invention adopts the characteristic cluster Ci of electrical appliances and the Euclidean distance method to judge the type of electrical appliances, and has a smaller amount of calculation. The actual test results in Table 2 show that the power consumption of this system is less than 1 watt (Pw=0.97), which has the characteristics of green energy saving and is easy to industrialize.

[0084] The second prior art only uses power change as the basis for judging the switching event. This type of method is only effective for high-power electrical appliances. From Table 1 and Figure 2 It can be seen that the applicable electrical power is more than tens of watts; the present invention adopts the parameter Vt which combines multiple changes such as power change, spectrum change, power factor change, etc. as the basis for judging the switching event, which has higher sensitivity. Figure 4 It can be seen from the simulation results and Table 1 that the present invention can realize the identification of small electrical appliances of about 1W; the second prior art adopts the Boot algorithm to calculate the entropy increase variable of the sampling sequence to realize electrical classification and identification; the present invention adopts the characteristic cluster Ci of electrical appliances and the Euclidean distance method to determine the type of electrical appliances; the two methods are obviously different.

[0085] The differences between the present invention and the prior art three are as follows:

[0086] (1) The feature parameter sets used for learning and classification are different. The prior art 3 is the volt-ampere ratio, harmonic distortion coefficient, and active power. The present invention is the power factor angle φ(k), the amplitude ratios D1 and D2 of the third and fifth harmonics to the fundamental wave, the current I of the electrical appliance, and the amplitude difference ΔD of the spectrum at 50 Hz. t (1);

[0087] (2) It can be seen from Tables 1 and 2 of the prior art 3 that the classification method of the prior art 3 is suitable for high-power electrical appliances (the minimum power electrical appliance in the table is a lithium battery: 0.291*220=64 watts), while the classification method of the present invention is suitable for high-power electrical appliances (the minimum power electrical appliance in the table is a lithium battery: 0.291*220=64 watts). Figure 5 It can be seen from the simulation results and Table 1 that the present invention can realize the recognition of small electrical appliances of about 1W (the power of 47K homemade electrical appliances is 0.005*220=1.1 watts, and the power of LED light strips is 9.9 watts);

[0088] (3) The prior art 3 uses KNN and VFKN algorithms to realize electrical appliance identification; the present invention uses the characteristic cluster Ci of electrical appliances and the Euclidean distance method to determine the type of electrical appliances; the two methods are obviously different.

[0089] Second, considering the technical solution as a whole or from the perspective of the product, the technical effects and advantages of the technical solution to be protected by the present invention are described in detail as follows:

[0090] The non-intrusive electrical appliance identification system provided by the present invention operates with a current of less than 1 watt and has low power consumption characteristics. After multiple experiments, it has been verified that the system of the present invention can successfully learn all electrical appliances to be learned in the learning mode and store the corresponding electrical parameters in the single-chip microcomputer; the system learns quickly, and the learning time of each electrical appliance is less than 1 minute. The algorithm and system proposed by the present invention have high practical value.

[0091] The present invention combines low cost, low power consumption and high recognition accuracy, and is easy to industrialize and promote. It has a learning function and can autonomously learn new electrical appliances. The electrical appliances that can be recognized are not limited to the types of electrical appliances built in at the factory. It only needs to train a small amount of data on a single electrical appliance to accurately recognize a combination of electrical appliances. BRIEF DESCRIPTION OF THE DRAWINGS

[0092] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments of the present invention. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0093] Figure 1 is a flow chart of a non-intrusive electrical appliance identification method provided by an embodiment of the present invention;

[0094] Figure 2 is a schematic diagram of a non-intrusive electrical appliance identification method provided by an embodiment of the present invention;

[0095] Figure 3 It is a hardware design diagram of a non-intrusive electrical appliance identification system provided by an embodiment of the present invention;

[0096] Figure 4 is an example electrical appliance spectrum diagram provided by an embodiment of the present invention;

[0097] Figure 5 is a schematic diagram of changes in an example electrical appliance provided by an embodiment of the present invention;

[0098] Figure 6 This is an OLED display diagram of a non-invasive electrical appliance identification system provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0099] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0100] In view of the problems existing in the prior art, the present invention provides a non-intrusive electrical appliance identification method, system, medium, device and terminal. The present invention is described in detail below with reference to the accompanying drawings.

[0101] 1. Explanatory Examples In order to enable those skilled in the art to fully understand how to implement the present invention, this section provides an illustrative example that expands and describes the technical solution of the claims.

[0102] Explanation of terms: Electrical appliance identification: refers to distinguishing and confirming the type of electrical appliances in use; non-invasive: refers to not changing the state of the original normal working system, having no impact on the normal working system, and existing in an additional state; dynamic characteristic parameters: refers to electrical parameters that change in real time, such as circuit current, electric power, etc.

[0103] The non-intrusive electrical appliance identification method provided in the embodiment of the present invention uses an MCU to process synchronously sampled current data of 500ms per group to obtain parameters such as the fundamental phase of the current, the amplitude ratio of each harmonic to the fundamental, and the energy ratio. These parameters are then used to calculate the characteristic quantities of each electrical appliance.

[0104] like Figure 1 As shown, the non-intrusive electrical appliance identification method provided by the embodiment of the present invention includes the following steps:

[0105] S101, synchronously sampling the current of the electrical appliance using a current transformer and a voltage zero-crossing detection circuit;

[0106] S102, using the MCU to process the synchronously sampled current data of 500 ms per group, and obtain the parameters of the fundamental phase of the current, the amplitude ratio of each harmonic to the fundamental, and the energy ratio;

[0107] S103, using current parameters to calculate the characteristic quantity of each electrical appliance; by extracting dynamic characteristic values ​​from the changing data, determining the type of the changing electrical appliance, and combining accumulation and identifying the type and quantity of the current electrical appliances.

[0108] The software system design scheme provided by the embodiment of the present invention is as follows Figure 2When the system starts, the OLED screen will display the system function menu. The matrix keyboard is used to select the system to enter the learning mode or detection mode. In the learning mode, the number of electrical appliances is 1. After the electrical appliances are enabled, the current waveforms and calculated characteristic parameters of each electrical appliance are collected and learned in turn through human-computer interaction, and these characteristic parameters are stored; in the monitoring mode, the system will periodically collect power consumption data and calculate the characteristic parameter V of power consumption. t , then according to V t If so, the dynamic feature parameters are calculated using the changed data and the Euclidean distance is used to determine the type of electrical appliances that have changed. Finally, the current type and quantity of electrical appliances identified are updated based on the cumulative sum (CUSUM) and displayed on the OLED screen.

[0109] The specific electrical appliance parameter processing technology process provided by the embodiment of the present invention is divided into electrical appliance parameter acquisition and electrical appliance change state extraction.

[0110] During the process of acquiring the parameters of electrical appliances, the MCU needs to read the current and voltage information synchronously collected by ADS1256, and then analyze and obtain useful data. In this system, the MCU reads the data of the ADC module through the serial peripheral interface (SPI). The sampling rate is set to 7.5K. The MCU interrupt is triggered by the DRDY signal of ADS1256 to realize continuous acquisition of signals. In order to ensure that the signals collected each time are in phase, the ADC is triggered by the rising edge of the zero-crossing signal to collect a set of data each time, with 4096 data in each set, and the circular FIFO queue is used to store the electrical parameter information. This method can effectively reduce the overhead of data processing and the interrupt service program of the DRDY signal.

[0111] In the process of extracting the state of the electrical appliance, the present invention realizes the extraction of the current spectrum characteristics through FFT operation. FFT is a fast algorithm of discrete Fourier transform (DFT), and the two have the same properties. Suppose the current sampling data of the tth group of samples is f t (l), then the result of the DFT of its N points is:

[0112]

[0113] Taking a table lamp, a hair dryer and a router as examples, the spectrum actually collected and calculated is as follows: Figure 4 shown.

[0114] Depend on Figure 4 It can be clearly seen that different electrical appliances have large differences in current spectra, so the current spectrum can be used as one of the bases for identifying electrical appliances.

[0115] In order to determine the number of electrical appliances and the time point of state change, it is necessary to study the instantaneous load characteristics of electrical appliances. Instantaneous load characteristics refer to the electrical characteristics of the load at the moment of opening and closing, and steady-state load characteristics refer to the electrical characteristics of the load after it is turned on and enters stable operation. The changes in the steady-state load characteristics of multiple electrical appliances can be used to determine the increase or decrease of electrical appliances.

[0116] The present invention defines that when multiple electrical appliances are working simultaneously, the values ​​of the 50 Hz fundamental frequency and each harmonic in the current spectrum collected are:

[0117]

[0118] Where fs is the sampling rate, Denotes floor rounding operation. Define Q t As an evaluation function:

[0119]

[0120] like Figure 5 As shown, at the moment t=6 when the electrical appliance changes, Q t There has been a significant change, namely, through Q t The present invention uses the change value ΔQ of the data of the interval m groups. t As a preliminary basis for judging whether electrical appliances have changed:

[0121]

[0122] Wherein H is a threshold, and the test of the present invention selects H=50mA, and m=2. The interval m can be adjusted according to the actual situation.

[0123] In order to further improve the recognition success rate of the increase and decrease of electrical appliances, the present invention adopts the differential method and selects the amplitude difference ΔD of the spectrum at 50Hz t (1) As additional judgment conditions:

[0124] ΔD t (1)=|D t (1)|-|D t-k (1)|;

[0125] By comparing ΔD t (1) Values ​​can quickly distinguish the increase or decrease of high-power electrical appliances.

[0126] Since electrical appliances may have different power factors, the present invention also selects the power factor angle φ t (k) As one of the characteristic parameters for identifying changes in electrical appliances:

[0127] φ t (k) = arg(Dt (k));

[0128] Here, arg(x) means the angle of x.

[0129] In summary, the function used to judge the change of electrical appliance status is:

[0130]

[0131] After processing the electrical appliance parameters, the present invention performs a specific electrical appliance identification process, and the detailed electrical appliance identification method is as follows:

[0132] 1. Define the ratio of the amplitude of each harmonic of the electrical current to the amplitude of the 50Hz fundamental frequency as

[0133]

[0134] make Eigenvector representing the harmonic ratios.

[0135] 2. Define the eigenvector Z t :

[0136]

[0137] in, is the effective value of current:

[0138]

[0139] 3. In the learning mode, multiple feature vectors Z1, Z2, Z3… are obtained by performing multiple recognitions on the features of a single electrical appliance as the feature cluster C of the electrical appliance. i :

[0140]

[0141] Among them, N c is the number of sample electrical appliances, N i is the number of learning times of the i-th electrical appliance.

[0142] 4. In the monitoring mode, obtain a characteristic vector d of a group of unknown samples in a steady state (the current electrical state remains unchanged):

[0143]

[0144] When a sample d is given, the feature cluster of electrical appliances is divided into C = {C1, C2, C3, ...}, and the minimum square error can be obtained:

[0145]

[0146] Among them, u i It is cluster C i The mean vector of :

[0147]

[0148] E describes the similarity between the sample and the cluster mean vector. The smaller the E value, the higher the similarity between the sample and the corresponding electrical appliance. By determining the value of the corresponding electrical appliance number i when the E value is the smallest, the type of the changed electrical appliance can be determined.

[0149] 5. After identifying the type of electrical appliances that have changed, according to ΔQ t Determine the state of the electrical appliances. t >0, the appliance is turned on, when ΔQ t <0, the appliance is turned off. Finally, according to the cumulative sum (CUSUM) principle, the type and quantity of the current appliance are updated, and the OLED display output is updated.

[0150] The non-intrusive electrical appliance identification system design provided by the embodiment of the present invention is divided into hardware system design and software system design. The hardware design scheme provided by the embodiment of the present invention is as follows: Figure 3 shown.

[0151] The MCU of the present invention adopts STM32H7A3 of STMicroelectronics, and the peripherals are composed of modules such as current sampling, voltage zero-crossing detection, power supply, ADC sampling, OLED display screen and keyboard.

[0152] The non-invasive electrical appliance identification system provided by the embodiment of the present invention first uses a current sampler to sample the electrical appliance current and uses a zero-crossing detection circuit to detect the time when the mains voltage crosses zero. Secondly, the ADC with a built-in PGA is used to amplify and convert the signal into analog-to-digital. Then, the MCU is used to synchronously collect the current signal at the time when the voltage crosses zero, intelligently detect the type and quantity of electrical appliances, and finally display the results on an OLED screen.

[0153] Among them, MCU can be other types of microprocessors with similar sampling and processing functions; specific parameters in the evaluation function and the function for judging the change of the state of the electrical appliance can select other values; the number and type of electrical appliance parameters used in the judgment process (Nth harmonic and fundamental wave ratio, power factor angle, etc.) can be modified and adjusted according to actual identification needs and cost considerations. The specific screen of the display hardware part of the present invention can be replaced, such as the HMI screen, LCD screen, etc., and the display part can also be changed according to actual needs.

[0154] 2. Application Examples: In order to prove the creativity and technical value of the technical solution of the present invention, this section provides application examples of the technical solution of the claims on specific products or related technologies.

[0155] The inventor first deployed the system implemented by the present invention on the input line of a wiring board, and connected a plurality of electrical appliance combinations to the wiring board, completed learning and identification, and verified the correctness of the system; then, the system implemented by the present invention was deployed in the electric meter box of the EII-202 laboratory of Xidian University, and a test showed that the system can accurately identify the combination of electrical appliances such as oscilloscopes, signal sources, power supplies, soldering irons, air conditioners, notebooks, desk lamps, etc. used in the laboratory, further verifying the correctness of the system.

[0156] 3. Evidence of the effects of the embodiments. The embodiments of the present invention have achieved some positive effects during the development or use process, and indeed have great advantages over the prior art. The following content is described in conjunction with the data, charts, etc. of the test process.

[0157] 1. Electrical appliance learning test

[0158] When learning, first connect the appliance to be learned to the socket and select to enter the learning mode through the human-computer interaction interface. After starting the learning, start the appliance and keep it running. At this time, the system will sample the electrical parameter information of a single appliance and calculate the characteristic vector Z of the appliance according to the detailed appliance identification method proposed. i After multiple identifications, the characteristic cluster C of the electrical appliance is obtained. i After successful learning, the learning results will be displayed on the OLED display screen, including the power factor angle φ(k), the amplitude ratios of the 3rd and 5th harmonics to the fundamental wave D1 and D2, the electrical current I and the system power consumption P w .

[0159] The results of the electrical appliance learning test are shown in Table 1.

[0160] Table 1. Learning mode test results

[0161]

[0162] As shown in Table 1, the system operating current is less than 1 watt, which has low power consumption characteristics. After multiple experiments, it was verified that the system can successfully learn all the electrical appliances to be learned in the learning mode and store the corresponding electrical parameters in the microcontroller. The system learns quickly, and the learning time of each electrical appliance is less than 1 minute.

[0163] 2. Electrical appliance identification test

[0164] The present invention verifies 256 different switching combinations of 8 kinds of electrical appliances, setting the on state of the electrical appliances to "1" and the off state to "0". The states of the various electrical appliances are represented by each bit of a byte, and electrical appliances No. 1 to 8 correspond to the 0th to 7th bits of the byte respectively. At this time, the 256 combinations of electrical appliances can be represented as a number from 0 to 255, which is represented in binary as {000000000, 00000001, 00000010, ..., 11111110, 11111111}. In order to eliminate the erroneous statistics caused by accumulated errors, each combination of electrical appliances starts with all electrical appliances being turned off, and the number of electrical appliances is gradually increased according to the second column of Table 2, and then the system recognition results are recorded in the third column of Table 2.

[0165] Table 2 Electrical appliance identification test results

[0166]

[0167] Table 2 gives the test results of different combinations of electrical appliances, and the omitted test results are all correct. It can be seen from Table 2 that errors occurred in the system during the 49th, 170th, and 237th experiments. Further analysis shows that when the error occurred, the No. 6 high-power electrical appliance (hair dryer) was working, resulting in the identification error of the smallest electrical appliance No. 5 (47k pure resistance). The recognition success rate of the system is (256-3) / 256=98.83%. Considering that the purpose of electrical appliance identification is to count the power load information, for such very small electrical appliances, the wrong recognition results have little effect on the statistics of the power load information. Therefore, the algorithm and system proposed in the present invention still have high practical value.

[0168] Figure 6 The OLED display information during system learning and testing is given. Figure 6 The left picture shows the results of the system learning energy-saving lamps, the middle picture shows the results when the system learns fans, and the right picture shows the results when the system detects that the LED light strip is turned on in monitoring mode. Figure 6 In the table, the number X in DATA X and IDX corresponds to the number of the electrical appliance in Table 1.

[0169] It should be noted that the embodiments of the present invention can be implemented by hardware, software, or a combination of software and hardware. The hardware part can be implemented using dedicated logic; the software part can be stored in a memory and executed by an appropriate instruction execution system, such as a microprocessor or dedicated design hardware. It can be understood by a person of ordinary skill in the art that the above-mentioned devices and methods can be implemented using computer executable instructions and / or contained in a processor control code, such as a carrier medium such as a disk, CD or DVD-ROM, a programmable memory such as a read-only memory (firmware), or a data carrier such as an optical or electronic signal carrier. Such code is provided on the carrier medium. The device and its modules of the present invention can be implemented by hardware circuits such as very large-scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, etc., or programmable hardware devices such as field programmable gate arrays, programmable logic devices, etc., can also be implemented by software executed by various types of processors, and can also be implemented by a combination of the above-mentioned hardware circuits and software, such as firmware.

[0170] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any modifications, equivalent substitutions and improvements made by any technician familiar with the technical field within the technical scope disclosed by the present invention and within the spirit and principle of the present invention should be covered by the protection scope of the present invention.

Claims

1. A non-intrusive electrical appliance identification method, characterized in that: The non-intrusive electrical appliance identification method comprises: The current of electrical appliances is synchronously sampled using a current transformer and a voltage zero-crossing detection circuit; the synchronously sampled current data of 500ms is processed by an MCU to obtain the parameters of the fundamental phase of the current, the amplitude ratio of each harmonic to the fundamental, and the energy ratio, and the characteristic quantities of each electrical appliance are calculated using the current parameters; the type of electrical appliance that has changed is determined by extracting dynamic characteristic values ​​from the changing data, and the type and quantity of the current electrical appliances are intelligently identified by combining accumulation and CUSUM; The electrical appliance parameter processing technology process in the non-invasive electrical appliance identification method is divided into electrical appliance parameter acquisition and electrical appliance change state extraction; During the process of acquiring the parameters of electrical appliances, the MCU reads the current and voltage information synchronously collected by ADS1256, and then analyzes and obtains useful data; the MCU reads the data of the ADC module through the serial peripheral interface SPI; the sampling rate is set to 7.5K, and the MCU interrupt is triggered by the DRDY signal of ADS1256 to achieve continuous acquisition of signals; each time the rising edge of the zero-crossing signal triggers the ADC to collect a set of data, and a circular FIFO queue is used to store the electrical parameter information; In the process of extracting the state of electrical appliances, the current spectrum characteristics are extracted through FFT operation. Suppose the current sampling data of the tth group is f t (k), then the result of DFT at N points is: The non-intrusive electrical appliance identification method further includes: When multiple electrical appliances are working at the same time, the 50Hz fundamental frequency and harmonics in the collected current spectrum are defined as: Where fs is the sampling rate, Represents the floor operation; define Q t As an evaluation function: The change value ΔQ of the data of the interval m group is used t As a preliminary basis for judging whether electrical appliances have changed: Among them, H is the threshold, select H = 50mA, m = 2; the interval m is adjusted according to the actual situation; The difference method is used, and the amplitude difference ΔD of the spectrum at 50Hz is selected. t (1) As additional judgment conditions: ΔD t (1)=|D t (1)|-|D t-k (1)|; By comparing ΔD t (1) The value distinguishes the increase or decrease of high-power electrical appliances; Select the power factor angle φ t (k) As one of the characteristic parameters for identifying changes in electrical appliances: φ t (k)=arg(D t (k)); Among them, arg(x) means to find the angle of x; The function used to judge the change of electrical appliance status is: After processing the electrical appliance parameters, the electrical appliance identification method is as follows: (1) Define the ratio of the amplitude of each harmonic of the electrical current to the amplitude of the 50 Hz fundamental frequency as make The eigenvector representing the harmonic ratio; (2) Define the eigenvector Z t : in, is the effective value of current: (3) In the learning mode, the characteristics of a single electrical appliance are identified multiple times to obtain multiple feature vectors Z1, Z2, Z3, etc., which are used as the feature cluster C of the electrical appliance. i : Among them, N c is the number of sample electrical appliances, N i is the number of learning times of the i-th electrical appliance; (4) In monitoring mode, the feature vector d of a set of unknown samples whose current electrical state remains unchanged is obtained: When a sample d is given, the minimum square error is obtained by dividing the characteristic clusters of electrical appliances into C = {C1, C2, C3, ...}: Among them, u i It is cluster C i The mean vector of : E describes the similarity between the sample and the cluster mean vector. The smaller the E value, the higher the similarity between the sample and the corresponding electrical appliance. By determining the value of the corresponding electrical appliance number i when the E value is the smallest, the type of the changed electrical appliance can be determined. (5) After identifying the type of electrical appliances that have changed, according to ΔQ t Determine the state of the electrical appliance; when ΔQ t >0, the electrical appliance is turned on, when ΔQ t <0, the electrical appliance is turned off; based on the accumulation and principle, the type and quantity of the current electrical appliance are updated, and the OLED display output is updated.

2. The non-intrusive electrical appliance identification method according to claim 1, characterized in that: The non-intrusive electrical appliance identification method comprises the following steps: Step 1: Initialize ADC and MCU; Step 2: Determine the current mode: determine whether to enter the learning mode; if entering the learning mode, enter the appliance number and execute step 3; if not entering the learning mode, enter the detection mode and execute step 5; Step 3, determine whether the learning key is pressed: if the learning key is pressed, data is collected synchronously and characteristic parameters are calculated and stored, prompting that the learning is completed and displaying the characteristic parameters; if the learning key is not pressed, execute step 4; Step 4: Determine whether to exit the learning mode: If exiting the learning mode, enter the detection mode and execute step 5; if not exiting the learning mode, return to step 3 and perform the learning operation again; Step 5, start the detection operation: collect data regularly, calculate the characteristic parameter Vt, and determine whether the change of Vt exceeds the threshold; if not, it means that the number and type of the currently identified electrical appliances have not changed, the identification result remains unchanged, and return to step 2; if yes, it means that the number and type of the currently identified electrical appliances have changed, and the identification result needs to be updated, and execute step 6; Step 6: Calculate dynamic feature parameters, use Euclidean distance to identify changing electrical appliances, and update the type and quantity of current electrical appliances based on the cumulative sum CUSUM.

3. The non-intrusive electrical appliance identification method according to claim 1, characterized in that: The non-intrusive electrical appliance identification method further includes: When the system starts, the OLED screen will display the system function menu. The matrix keyboard will be used to select the system to enter the learning mode or detection mode. In the learning mode, the number of electrical appliances is 1. After the electrical appliances are enabled, the current waveforms and calculated characteristic parameters of each electrical appliance are collected and learned in turn through human-computer interaction, and these characteristic parameters are stored. In the monitoring mode, the system will periodically collect power consumption data and calculate the characteristic parameter V of power consumption. t , and according to V t The change in the status of electrical appliances is used to determine whether there is a change in the status of electrical appliances. If so, the dynamic feature parameters are calculated using the changed data and the Euclidean distance is used to determine the type of electrical appliances that have changed; the current type and quantity of electrical appliances are identified based on accumulation and update and displayed on the OLED screen.

4. A non-intrusive electrical appliance identification system using the non-intrusive electrical appliance identification method according to any one of claims 1 to 3, characterized in that: The non-intrusive electrical appliance identification system includes an MCU; The MCU adopts STM32H7A3, and the peripherals are composed of a current sampling module, a voltage zero-crossing detection module, a power module, an ADC sampling module, an OLED display module and a keyboard module; The current sampler is used to sample the current of electrical appliances and the zero-crossing detection circuit is used to detect the moment when the mains voltage crosses zero; the ADC with built-in PGA is used to amplify and convert the signal into analog-to-digital, and the MCU is used to synchronously collect current signals and intelligently detect the types and quantities of electrical appliances at the moment when the voltage crosses zero, and the results are displayed on the OLED screen.

5. A computer device, characterized in that: The computer device includes a memory and a processor, the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the non-intrusive electrical appliance identification method according to any one of claims 1 to 3, comprising the following steps: The current of electrical appliances is synchronously sampled using a current transformer and a voltage zero-crossing detection circuit. The synchronously sampled current data of 500ms is processed by an MCU to obtain the parameters of the fundamental phase of the current, the amplitude ratio of each harmonic to the fundamental, and the energy ratio, and the characteristic quantity of each electrical appliance is calculated using the current parameters. The type of electrical appliance that has changed is determined by extracting dynamic characteristic values ​​from the changing data, and the type and quantity of the current electrical appliances are intelligently identified by combining accumulation and CUSUM.

6. A computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, the processor executes the non-intrusive electrical appliance identification method according to any one of claims 1 to 3, comprising the following steps: The current of electrical appliances is synchronously sampled using a current transformer and a voltage zero-crossing detection circuit. The synchronously sampled current data of 500ms is processed by an MCU to obtain the parameters of the fundamental phase of the current, the amplitude ratio of each harmonic to the fundamental, and the energy ratio, and the characteristic quantity of each electrical appliance is calculated using the current parameters. The type of electrical appliance that has changed is determined by extracting dynamic characteristic values ​​from the changing data, and the type and quantity of the current electrical appliances are intelligently identified by combining accumulation and CUSUM.

7. An information data processing terminal, characterized in that: The information data processing terminal is used to implement the non-intrusive electrical appliance identification system as claimed in claim 5.

Images