Load mode detection method and system based on accumulation and geometric feature analysis

Abstract

The invention discloses a load mode detection method and system based on accumulation and geometric feature analysis, and relates to the technical field of non-intrusive power load monitoring, and the method comprises the steps: carrying out the preprocessing of a current signal, outputting a current root-mean-square signal, and eliminating noise interference; judging a non-stationary point based on a current mean square sequence output by preprocessing and generating a corresponding CUSUM sequence; and detecting the load mode in the CUSUM sequence. According to the load mode detection method based on accumulation and geometric feature analysis provided by the invention, the accuracy and stability of the whole detection system are improved by preprocessing the current signal, the sudden change point in the load mode can be effectively identified by using the CUSUM sequence, the sensitivity and accuracy of load mode detection are improved, and the detection accuracy of the load mode is improved. The generated CUSUM sequence is analyzed, the accuracy and sensitivity of load event detection are improved, and better effects are achieved in the aspects of accuracy, reliability and efficiency.

Description

Technical Field

[0001] The present invention relates to the technical field of non-intrusive power load monitoring, and in particular to a load mode detection method and system based on accumulation and geometric feature analysis. Background Art

[0002] Power supply stability is a key issue in maintaining the balance between power supply and demand. In order to maintain the stability of power supply, many experts actively promote demand response policies, which are currently being implemented in many countries around the world to achieve a stable supply of electricity.

[0003] Non-intrusive load monitoring (NILM) was proposed in the 1980s and is an important technology widely used in demand response. The core purpose of NILM is to identify the power consumption behavior of individual appliances by analyzing the total power characteristics. NILM can collect power consumption information of appliances within the end-user and combine it with demand response measures to achieve a balanced and mutually beneficial balance of power supply and demand.

[0004] In recent years, with the continuous development of power systems and the increase in power load demand, higher requirements have been placed on the detection technology of power load patterns. Traditional load pattern detection methods usually rely on simple statistical analysis and empirical rules, which have shown certain limitations when facing complex and changeable power load characteristics. In order to improve the accuracy and efficiency of detection, modern power systems have gradually introduced advanced technologies based on signal processing and big data analysis. These technologies can more accurately identify and predict changes in load patterns through in-depth analysis of power signals, thereby providing important support for the optimized operation of power systems.

[0005] Existing load mode detection methods mainly include technologies based on time series analysis, frequency domain analysis and wavelet transform. However, these methods often find it difficult to maintain high detection accuracy when facing non-stationary signals. Especially in the case of large noise interference, traditional methods find it difficult to effectively distinguish useful signals from noise, resulting in a high false detection rate. In addition, existing methods often lack unified standards and systematic theoretical support in the preprocessing and feature extraction of current signals. Therefore, developing a method that can efficiently and accurately detect load modes in complex environments has become a hot topic and difficulty in current research. Summary of the invention

[0006] In view of the above-mentioned problems, the present invention is proposed.

[0007] Therefore, the technical problem solved by the present invention is that the existing load mode detection method has the problems of low detection accuracy, high false detection rate, low efficiency, and difficulty in effectively distinguishing useful signals from noise.

[0008] To solve the above technical problems, the present invention provides the following technical solutions: a load mode detection method based on accumulation and geometric feature analysis, comprising preprocessing the current signal to output a current root mean square signal and eliminating noise interference; determining non-stationary points based on the current root mean square sequence output by the preprocessing and generating a corresponding CUSUM sequence; and detecting the load mode in the CUSUM sequence.

[0009] As a preferred solution of the load mode detection method based on accumulation and geometric feature analysis described in the present invention, wherein: the preprocessing of the current signal to output the current RMS signal and eliminate noise interference includes preprocessing the RMS value of each cycle of the output total current signal, outputting the current RMS sequence, and eliminating noise data through noise geometric features to complete noise reduction;

[0010] The current signal preprocessing takes the total current signal collected as input and outputs the current root mean square I for each current cycle. RMS It is expressed as:

[0011]

[0012] Among them, I i represents the mth current sampling value in the current current cycle, and M represents the number of sampling points in the current current cycle.

[0013] As a preferred solution of the load mode detection method based on accumulation and geometric feature analysis described in the present invention, wherein: the preprocessing of the current signal to output the current root mean square signal and eliminate noise interference includes outputting the current root mean square value for each current cycle, outputting the current root mean square sequence {I RMS (1),I RMS (2),...,I RMS (N)};

[0014] For the acquired current mean square sequence, a noise discrimination mechanism is defined to exclude noise points in the sequence to achieve noise reduction effect;

[0015] For each current mean square point I belonging to the current mean square sequence RMS (n)∈{I RMS (1),I RMS (2),...,I RMS (N)}, output first-order derivative sequence I RMS '(k)∈{I RMS '(1),I RMS '(2),…,I RMS '(N-1)};

[0016] The noise point is detected to build constraints, and the current mean square point is a noise point expressed as:

[0017] |I RMS '(k)|≥I th '∩|I RMS '(k+1)|≥I th ',1≤k≤N-1

[0018] Among them, I th ' represents the current mean square derivative threshold;

[0019] The noise point is detected to build constraints, and the current mean square point is a noise point expressed as:

[0020]

[0021] Among them, tp represents the position of the extreme point in the first-order derivative sequence, w th represents the noise observation window length and satisfies 1 <tp-w th / 2 <tp<tp+w th / 2<N,I th1 Indicates the current mean square value change judgment threshold;

[0022] If there is a current RMS point, I RMS If (n') satisfies one of the two constraints, it will be judged as a noise point, and the current mean square value corresponding to the noise point will be output to complete the noise reduction, which is expressed as:

[0023]

[0024] Among them, w p represents the predetermined noise suppression window length;

[0025] Perform noise reduction operations on all noise points and update the current mean square sequence.

[0026] As a preferred solution of the load mode detection method based on accumulation and geometric feature analysis described in the present invention, wherein: the determination of the non-stationary point based on the current mean square sequence output by the preprocessing and the generation of the corresponding CUSUM sequence includes combining the single point comparison and the sliding window comparison for the current mean square sequence obtained after the preprocessing to determine the non-stationary point, and performing CUSUM analysis on the non-stationary point to generate the CUSUM sequence;

[0027] Compare each current mean square point I in the current mean square sequence one by one RMS , if the current mean square point stability condition is met, then this current mean square point is determined as a stable point. The stability condition is expressed as:

[0028] I RMS ≤I th2

[0029] Among them, I th2 Indicates the current mean square stability threshold;

[0030] Set the number of elements to N w The sliding window long set Each element in the sliding window length set is used as a window length scanning current mean square sequence;

[0031] In each scanning process, if the current mean square value satisfied by the elements in the sliding window is compared with the preset threshold, if the current mean point stability condition in the sliding window is satisfied, the current mean point in the sliding window is determined to be a stable point:

[0032]

[0033] Among them, I max and I min Respectively represent the maximum and minimum current mean square value in the sliding window, I th3 Indicates the sliding window decision threshold.

[0034] As a preferred solution of the load mode detection method based on accumulation and geometric feature analysis described in the present invention, wherein: the determination of non-stationary points based on the current mean square sequence output by the preprocessing and the generation of the corresponding CUSUM sequence include identifying the current mean square point between every two non-adjacent stable points as a non-stationary sequence;

[0035] For a non-stable sequence with U elements {I RMS (u 1 ),I RMS (u 2 ),…,I RMS (U)} Perform CUSUM analysis and obtain the CUSUM sequence {C u1 ,C u2 ,…C U}, expressed as:

[0036] C u1 =(I RMS (u 1 )-μ)

[0037] C u2 =(I RMS (u 2 )-μ)+C u1

[0038] C u3 =(I RMS (u 3 )-μ)+C u2

[0039] …

[0040] C U =(I RMS (U)-μ)+CU-1

[0041] Where μ represents the unstable sequence {I RMS (u 1 ),I RMS (u 2 ),…,I RMS (U)} is the mean of all elements in .

[0042] As a preferred solution of the load mode detection method based on accumulation and geometric feature analysis described in the present invention, wherein: the detection of the load mode in the CUSUM sequence includes the detection of the power load mode on the output CUSUM sequence;

[0043] For each CUSUM sequence, find and record the local extreme points that meet the requirements, expressed as:

[0044]

[0045] Among them, m is the specified window length for determining the local extreme point;

[0046] It is preliminarily determined that there is a corresponding load pattern at the location of each extreme point;

[0047] If every two adjacent local extreme points with opposite polarity follow the output slope Slop extremum , if the slope Slop extremum If the load event existence condition is met, it is considered that there are load events at these two local extreme points, which can be expressed as:

[0048] Slop extremum ≥Slop th

[0049] Among them, Slop th Indicates the slope threshold.

[0050] As a preferred solution of the load pattern detection method based on cumulative sum geometric feature analysis described in the present invention, wherein: the detection of the load pattern in the CUSUM sequence includes determining whether the sample detected as the load pattern is a repetitive load pattern;

[0051] Output each sample of the output being judged, the maximum value of the first-order derivative of the mean square of the current before and after, and the difference T between the time when the maximum output value appears and the time when the load mode sample appears interval , if the repetitive load mode condition is met, it is determined to be a repetitive load mode, which is expressed as:

[0052] T interval =T th

[0053] Among them, T threpresents the repeatability determination threshold;

[0054] All samples that are judged to be repetitive load patterns are extracted and the final load pattern detection results are output.

[0055] Another object of the present invention is to provide a load mode detection system based on accumulation and geometric feature analysis, which can generate a current root mean square signal by preprocessing the current signal and eliminate noise interference. By eliminating noise interference, a purer current signal can be obtained, which lays the foundation for subsequent load mode detection, improves the accuracy and stability of the overall detection system, and solves the problem of low detection accuracy in current load mode detection methods.

[0056] As a preferred solution of the load mode detection system based on accumulation and geometric feature analysis described in the present invention, it includes: a preprocessing module, a sequence output module, and a load detection module; the preprocessing module is used to preprocess the current signal to generate a current root mean square signal; the sequence output module is used to determine the non-stationary point through the current root mean square sequence and generate the corresponding CUSUM sequence; the load detection module is used to formulate geometric features to detect the load mode from the CUSUM sequence.

[0057] A computer device comprises a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of a load mode detection method based on accumulation and geometric feature analysis.

[0058] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of a load mode detection method based on accumulation and geometric feature analysis.

[0059] Beneficial effects of the present invention: The load mode detection method based on accumulation and geometric feature analysis provided by the present invention preprocesses the current signal to generate a current root mean square signal, and eliminates noise interference, thereby improving the accuracy and stability of the overall detection system, determining non-stationary points from the preprocessed current root mean square sequence, and generating a corresponding CUSUM sequence. The CUSUM sequence can be used to effectively identify the mutation points in the load mode. By analyzing the current root mean square sequence, the non-stationary points in the current signal can be accurately located, and a CUSUM sequence that is helpful for detection can be generated, thereby improving the sensitivity and accuracy of load mode detection. The generated CUSUM sequence is used to detect the load mode, making the detection of the load mode more efficient and accurate, which is helpful for real-time monitoring of changes in power loads, optimizing the operation and management of power systems, calculating the current root mean square value for each current cycle, and eliminating noise data. , generate a current root mean square sequence after noise reduction, improve the purity of the current signal, reduce the impact of noise interference on the detection results, ensure the accuracy of load mode detection, establish a noise discrimination mechanism, identify and exclude noise points in the current root mean square sequence, and further improve the quality of the current signal through precise noise reduction operations, providing a guarantee for the accurate detection of the load mode. Combined with single-point comparison and sliding window comparison, the unstable point is determined and a CUSUM sequence is generated to improve the recognition accuracy of the unstable point. The generated CUSUM sequence is more effective, and the accuracy and reliability of load mode detection are enhanced. The generated CUSUM sequence is analyzed to find and record local extreme points that meet the conditions, improve the accuracy and sensitivity of load event detection, and can better monitor and manage power loads and optimize the allocation and use of power resources. The present invention achieves better results in terms of accuracy, reliability and efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0061] Figure 1 An overall flow chart of a load mode detection method based on accumulation and geometric feature analysis provided in the first embodiment of the present invention.

[0062] Figure 2 A diagram illustrating a load pattern detection method based on accumulation and geometric feature analysis provided in the first embodiment of the present invention.

[0063] Figure 3 An overall flow chart of a load mode detection system based on accumulation and geometric feature analysis provided in the third embodiment of the present invention. DETAILED DESCRIPTION

[0064] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0065] Example 1, reference Figure 1-Figure 2 , as an embodiment of the present invention, provides a load mode detection method based on accumulation and geometric feature analysis, comprising:

[0066] S1: Pre-process the current signal to output the current RMS signal and eliminate noise interference.

[0067] Furthermore, the preprocessing module calculates the current root mean square value of each cycle of the total current signal, generates a current root mean square sequence, and eliminates noise data through noise geometric characteristics to achieve noise reduction.

[0068] It should be noted that the current signal preprocessing module takes the collected total current signal as input and calculates the current root mean square I for each current cycle as follows: RMS :

[0069]

[0070] Among them, I i represents the mth current sampling value in the current current cycle, and M represents the number of sampling points in the current current cycle. The current mean square value is calculated for each current cycle to obtain the current mean square sequence corresponding to N current cycles {I RMS (1),I RMS (2),…,I RMS (N)}.

[0071] For the acquired current mean square sequence, a noise discrimination mechanism is defined to exclude noise points in the sequence to achieve noise reduction effect. Specifically, for each current mean square point I in the current mean square sequence, RMS (n)∈{I RMS (1),I RMS (2),…,I RMS (N)}, calculate the first-order derivative sequence I RMS '(k)∈{I RMS '(1),I RMS '(2),…,I RMS '(N-1)}. The following constraints are proposed to detect noise points:

[0072] Constraint 1: The current mean square point that satisfies the following is a noise point:

[0073] |I RMS '(k)|≥I th '∩|I RMS '(k+1)|≥I th ',1≤k≤N-1

[0074] Among them, I th ' is the current mean square derivative threshold.

[0075] Constraint 2: The current mean square point that satisfies the following is a noise point:

[0076]

[0077] Among them, tp is the position of the extreme point in the first-order derivative sequence, w th is the noise observation window length, and satisfies 1 <tp-w th / 2 <tp<tp+w th / 2<N,I th1 It is the current mean square value change judgment threshold.

[0078] If a current RMS point, such as I RMS If (n') satisfies one of the above two constraints, it is determined as a noise point, and the current mean square value corresponding to the noise point is calculated as follows to complete the noise reduction:

[0079]

[0080] Among them, w p is the predetermined noise suppression window length. Perform noise reduction operations on all noise points and update the current mean square sequence.

[0081] S2: Determine the non-stationary point based on the current mean square sequence output by the preprocessing and generate the corresponding CUSUM sequence.

[0082] Furthermore, the current mean square sequence obtained after preprocessing is combined with single point comparison and sliding window comparison to determine the unstable point, and CUSUM analysis is performed on the unstable point to generate a CUSUM sequence.

[0083] It should be noted that it is determined that there may be an unstable point of the load mode in the current mean square sequence obtained in S1, and a corresponding CUSUM sequence is generated, as follows:

[0084] 1) Compare each current mean square point I in the current mean square sequence one by one RMS , if the following conditions are met, this current RMS point is determined as a stable point:

[0085] I RMS≤I th2

[0086] Among them, I th2 is the current mean square stability threshold.

[0087] 2) Set the number of elements to N w The sliding window long set Each element in the sliding window length set is used as a window length scanning current mean square sequence. In each scanning process, if the current mean square value satisfied by the element in the sliding window is compared with the preset threshold, if the following conditions are met, the current mean point in the sliding window is determined to be a stable point:

[0088]

[0089] Among them, I max and I min are the maximum and minimum current mean square values ​​in the sliding window, I th3 Indicates the sliding window decision threshold.

[0090] 3) The current mean square point between every two non-adjacent stable points is identified as an unstable sequence. RMS (u 1 ),I RMS (u 2 ),...,I RMS (U)} Perform the CUSUM analysis shown below to obtain the CUSUM sequence {C u1 ,C u2 ,…C U}:

[0091] C u1 =(I RMS (u 1 )-μ)

[0092] C u2 =(I RMS (u 2 )-μ)+C u1

[0093] C u3 =(I RMS (u 3 )-μ)+C u2

[0094] …

[0095] C U =(I RMS (U)-μ)+C U-1

[0096] Where μ represents the unstable sequence {I RMS (u1 ),I RMS (u 2 ),...,I RMS (U)} is the mean of all elements in .

[0097] S3: Detect the load pattern through the CUSUM sequence.

[0098] Furthermore, geometric features are formulated to detect load patterns from CUSUM sequences, and unreasonable load patterns are eliminated by analyzing the changing trends of CUSUM sequences.

[0099] It should be noted that the power load pattern detection of the CUSUM sequence obtained by S2 specifically includes:

[0100] For each CUSUM sequence, find and record local extreme points that meet the following requirements:

[0101]

[0102] Where m is the specified window length for determining the local extreme point. It is preliminarily determined that there is a corresponding load pattern at the location of each extreme point.

[0103] If every two adjacent local extreme points with opposite polarity are calculated according to the slope Slop extremum , if the slope Slop extremum If the following conditions are met, it is considered that there are load events at both local extreme points:

[0104] Slop extremum ≥Slop th

[0105] Among them, Slop th is the slope threshold.

[0106] Finally, determine whether the sample detected as a load pattern is a repetitive load pattern. Calculate the maximum value of the first-order derivative of the mean square of the current before and after each sample determined by the previous step, and calculate the difference T between the time when the maximum value appears and the time when the load pattern sample appears interval , if the following conditions are met, it is determined to be a repetitive load mode:

[0107] T interval =T th

[0108] Among them, T th is the repeatability judgment threshold. All samples judged as repeatable load patterns are extracted to obtain the final load pattern detection result.

[0109] Example 2, an embodiment of the present invention, provides a load mode detection method based on accumulation and geometric feature analysis. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments.

[0110] First, the COOLL and PLAID datasets with high-frequency data are used for testing. The evaluation index is the commonly used state identification accuracy index in the field of non-intrusive load monitoring, Accuracy.

[0111] In the embodiment, the key parameters are set as follows: current mean square derivative threshold I th 'Set to 3, the noise observation window length is w th Set to 5s, the noise suppression window length is w p Set to 3s, the current mean square stability threshold I th2 Set to 1e-3A, the sliding window judgment threshold I th3 is set to 0.7, the specified window length m for determining the local extreme point is 1s, and the slope threshold Slop th Set to 2, the repeatability threshold T th Set to 2.5s.

[0112] According to the above experimental configuration, training and testing are carried out, and the test results are shown in Table 1.

[0113] Table 1 Experimental data table

[0114]

[0115]

[0116] It can be seen from the results that the detection method of the present invention has significantly better identification results for various electrical appliances than the field of non-invasive load monitoring, and has excellent performance in some common electrical appliances, such as wall breakers, hair dryers, routers and other equipment.

[0117] In summary, the present invention achieves better results in terms of accuracy, reliability and efficiency.

[0118] Example 3, reference Figure 3 , as an embodiment of the present invention, provides a load pattern detection system based on accumulation and geometric feature analysis, including a preprocessing module, a sequence output module, and a load detection module.

[0119] The preprocessing module is used to preprocess the current signal and generate the current root mean square signal. The sequence output module is used to determine the non-stationary point through the current root mean square sequence and generate the corresponding CUSUM sequence. The load detection module is used to formulate geometric features to detect the load mode from the CUSUM sequence.

[0120] If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods of each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0121] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0122] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering, or processing in another suitable manner as necessary, and then stored in a computer memory.

[0123] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc. It should be noted that the above embodiments are only used to illustrate the technical solution of the present invention and are not limited. Although the present invention is described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention, which should be included in the scope of the claims of the present invention.

[0124] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A load mode detection method based on accumulation and geometric feature analysis, characterized in that: include: Pre-process the current signal to output the current root mean square signal and eliminate noise interference; Determine the non-stationary point based on the current mean square sequence output by the preprocessing and generate the corresponding CUSUM sequence; Detect load patterns in CUSUM sequences.

2. The load mode detection method based on accumulation and geometric feature analysis according to claim 1, characterized in that: The preprocessing of the current signal to output a current RMS signal and eliminating noise interference includes preprocessing the RMS value of each cycle of the output total current signal, outputting a current RMS sequence, and eliminating noise data through noise geometric features to achieve noise reduction; The current signal preprocessing takes the total current signal collected as input and outputs the current root mean square I for each current cycle. RMS It is expressed as: Among them, I i represents the mth current sampling value in the current current cycle, and M represents the number of sampling points in the current current cycle.

3. The load mode detection method based on accumulation and geometric feature analysis as claimed in claim 2, characterized in that: The preprocessing of the current signal to output a current root mean square signal and eliminate noise interference includes outputting a current root mean square value for each current cycle, outputting a current root mean square sequence {I RMS (1),I RMS (2),...,I RMS (N)}; For the acquired current mean square sequence, a noise discrimination mechanism is defined to exclude noise points in the sequence to achieve noise reduction effect; For each current mean square point I belonging to the current mean square sequence RMS (n)∈{I RMS (1),I RMS (2),...,I RMS (N)}, output first-order derivative sequence I RMS '(k)∈{I RMS '(1),I RMS '(2),...,I RMS '(N-1)}; The noise point is detected to build constraints, and the current mean square point is a noise point expressed as: |I RMS '(k)|≥I th '∩|I RMS '(k+1)|≥I th ',1≤k≤N-1 Among them, I th ' represents the current mean square derivative threshold; The noise point is detected to build constraints, and the current mean square point is a noise point expressed as: Among them, tp represents the position of the extreme point in the first-order derivative sequence, w th represents the noise observation window length and satisfies 1 <tp-w th / 2 <tp<tp+w th / 2<N,I th1 Indicates the current mean square value change judgment threshold; If there is a current RMS point, I RMS If (n') satisfies one of the two constraints, it will be judged as a noise point, and the current mean square value corresponding to the noise point will be output to complete the noise reduction, which is expressed as: Among them, w p represents the predetermined noise suppression window length; Perform noise reduction operations on all noise points and update the current mean square sequence.

4. The load mode detection method based on accumulation and geometric feature analysis as claimed in claim 3 is characterized in that: The method of determining the non-stationary point based on the current mean square sequence output by the preprocessing and generating the corresponding CUSUM sequence includes determining the non-stationary point by combining the single point comparison and the sliding window comparison with the current mean square sequence obtained after the preprocessing, and performing CUSUM analysis on the non-stationary point to generate the CUSUM sequence; Compare each current mean square point I in the current mean square sequence one by one RMS , if the current mean square point stability condition is met, then this current mean square point is determined as a stable point. The stability condition is expressed as: I RMS ≤I th2 Among them, I th2 Indicates the current mean square stability threshold; Set the number of elements to N w The sliding window long set Each element in the sliding window length set is used as a window length scanning current mean square sequence; In each scanning process, if the current mean square value satisfied by the elements in the sliding window is compared with the preset threshold, if the current mean point stability condition in the sliding window is satisfied, the current mean point in the sliding window is determined to be a stable point: Among them, I max and I min Respectively represent the maximum and minimum current mean square value in the sliding window, I th3 Indicates the sliding window decision threshold.

5. The load mode detection method based on accumulation and geometric feature analysis as claimed in claim 4, characterized in that: The method of determining the non-stationary point based on the current mean square sequence output by the preprocessing and generating the corresponding CUSUM sequence includes identifying the current mean square point between every two non-adjacent stable points as a non-stationary sequence; For a non-stable sequence with U elements {I RMS (u1),I RMS (u2),...,I RMS (U)} Perform CUSUM analysis and obtain the CUSUM sequence {C u1 ,C u2 ,...C U }, expressed as: C u1 =(I RMS (u1)-μ) C u2 =(I RMS (u2)-μ)+C u1 C u3 =(I RMS (u3)-μ)+C u2 … C U =(I RMS (U)-μ)+C U-1 Where μ represents the unstable sequence {I RMS (u1),I RMS (u2),...,I RMS (U)} is the mean of all elements in .

6. The load mode detection method based on accumulation and geometric feature analysis as claimed in claim 5, characterized in that: The detecting the load mode in the CUSUM sequence includes performing power load mode detection on the output CUSUM sequence; For each CUSUM sequence, find and record the local extreme points that meet the requirements, expressed as: Among them, m is the specified window length for determining the local extreme point; It is preliminarily determined that there is a corresponding load pattern at the location of each extreme point; If every two adjacent local extreme points with opposite polarity follow the output slope Slop extremum , if the slope Slop extremum If the load event existence condition is met, it is considered that there are load events at these two local extreme points, which can be expressed as: Slop extremum ≥Slop th Among them, Slop th Indicates the slope threshold.

7. The load mode detection method based on accumulation and geometric feature analysis according to claim 6 is characterized in that: The detecting the load pattern in the CUSUM sequence includes determining whether the sample detected as the load pattern is a repetitive load pattern; Output each sample of the output being judged, the maximum value of the first-order derivative of the mean square of the current before and after, and the difference T between the time when the maximum output value appears and the time when the load mode sample appears interval , if the repetitive load mode condition is met, it is determined to be a repetitive load mode, which is expressed as: T interval =T th Among them, T th represents the repeatability determination threshold; All samples that are judged to be repetitive load patterns are extracted and the final load pattern detection results are output.

8. A system using the load mode detection method based on accumulation and geometric feature analysis as claimed in any one of claims 1 to 7, characterized in that: Including pre-processing module, sequence output module, load detection module; The preprocessing module is used to preprocess the current signal to generate a current root mean square signal; The sequence output module is used to determine the non-stationary point through the current mean square sequence and generate the corresponding CUSUM sequence; The load detection module is used to develop geometric features to detect load patterns from CUSUM sequences.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the load mode detection method based on accumulation and geometric feature analysis according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the load mode detection method based on accumulation and geometric feature analysis according to any one of claims 1 to 7 are implemented.

Images