A method and device for extracting motor load events based on optimized sliding mean

Abstract

The present invention discloses a method and device for extracting motor load events based on sliding mean optimization. The method includes obtaining an active power sequence with a set sampling frequency at a pre-collected general industrial and commercial meter to be detected; dividing the active power sequence into multiple data windows by using a sliding window with a set size, and respectively calculating the mean values of the active power sequences in each data window; comparing the magnitudes of the mean values of the active power sequences in each data window with those in the adjacent previous data window. If the mean value of the active power sequence in the current data window is larger than that in the previous data window, an accumulation sequence is obtained through an accumulation sum algorithm; calculating the average slope difference of the accumulation sequence to obtain the forward difference of the accumulation sequence; determining the start-stop information of the motor load according to the forward difference of the accumulation sequence. The present invention is a non-intrusive method. By only collecting data at the user's general meter, the monitoring effect can be achieved. The number of devices used is small, the cost is low, and it is easy to implement, providing an important basis for government supervision.

Description

Technical Field

[0001] The present invention relates to a method and device for extracting motor load events optimized based on a sliding mean, belonging to the technical field of power research. Background Art

[0002] In 2019, with the occurrence of multiple chemical plant explosion incidents across the country, the government has continuously increased its supervision of high-risk industrial and commercial users. However, some industrial and commercial users, such as illegal "small chemical plants" with high fire risks and large environmental pollution hazards, hide in remote areas and secret corners, posing great difficulties to the government's safety supervision and environmental protection supervision, and also putting forward higher requirements for real-time and accurate identification of industrial and commercial loads.

[0003] To cope with the increasingly prominent power demand gap and safety supervision problems, the research on power consumption behavior monitoring technology for general industrial and commercial users is particularly important. General industrial and commercial users have complex types, a wide variety of load types and diverse application scenarios, and the load identification algorithms on the residential side are difficult to directly apply. Moreover, the current research on general industrial and commercial load identification has just started, and the overall research level still remains at the theoretical research stage, urgently needing to achieve a technological breakthrough. At the same time, there are a large number of large-fluctuation motor loads in industry and commerce, which will generate a large amount of power changes during operation, and ordinary load identification algorithms are difficult to identify. Summary of the Invention

[0004] The purpose of the present invention is to overcome the deficiencies in the prior art and provide a method and device for extracting motor load events optimized based on a sliding mean to solve the above problems.

[0005] To achieve the above purpose, the present invention is implemented by the following technical solutions:

[0006] In the first aspect, the present invention provides a method for extracting motor load events optimized based on a sliding mean, including:

[0007] Obtain the active power sequence with a set sampling frequency at the pre-collected general meter of the industrial and commercial users to be detected;

[0008] Use a sliding window with a set size to divide the active power sequence into multiple data windows, and calculate the mean value of the active power sequence in each data window respectively;

[0009] Compare the mean value of the active power sequence of each data window with that of the adjacent previous data window. If the mean value of the active power sequence of the current data window is larger than that of the previous data window, obtain an accumulation sequence through an accumulation sum algorithm;

[0010] Calculate the average slope difference of the accumulation sequence to obtain the forward difference of the accumulation sequence;

[0011] Determine the start-stop information of motor loads according to the forward difference of the accumulation sequence.

[0012] Further, it also includes: after obtaining the active power sequence, select a fixed window, and bring the active power sequence in the fixed window into the Gaussian function for smoothing processing. The formula is as follows:

[0013]

[0014] Among them, f k is the signal after smoothing processing, x is the active power sequence, σ is the variance in a window, and μ is the mean value of the active power sequence x with every multiple points as a window.

[0015] Further, divide the active power sequence into multiple data windows by using a sliding window of a set size, and calculate the mean value of the active power sequence in each data window respectively. The formula is as follows:

[0016]

[0017] Among them, K n is the numerical mean value of the active power sequence x with subscripts from n to n + m - 1; m is the window size.

[0018] Further, compare the mean value of the active power sequence in each data window with that in the adjacent previous data window. If the mean value of the active power sequence in the current data window is larger than that in the previous data window, obtain the accumulation sequence Z n , and the formula is as follows:

[0019] Z n = max(0, Z n-1 + x n -(K n - K n-1 ))

[0020] Further, calculate the average slope difference of the accumulation sequence to obtain the forward difference of the accumulation sequence. The formula is as follows:

[0021]

[0022] P d = P r - P r-1

[0023] Among them: P r is the mean value of the sequence Z with subscripts from n to n + m - 1, and P d is the forward difference of the sequence P with subscript r;

[0024] Further, determining the start-stop information of motor loads according to the forward difference of the accumulation sequence includes:

[0025] Compare the value of P d with a pre-set threshold value. After exceeding a certain threshold value, it is determined that an event occurs at this time, and the start-stop information of the motor-type load is obtained.

[0026] In a second aspect, the present invention provides a motor-type load event extraction device based on sliding mean optimization, including:

[0027] An acquisition unit for acquiring the active power sequence with a set sampling frequency at the total industrial and commercial meter to be detected collected in advance;

[0028] A mean calculation unit for dividing the active power sequence into multiple data windows by using a sliding window of a set size, and respectively calculating the mean values of the active power sequences in each data window;

[0029] A judgment unit for comparing the mean values of the active power sequences of each data window with those of the adjacent previous data window. If the mean value of the active power sequence of the current data window is larger than that of the previous data window, an accumulation sequence is obtained through the accumulation sum algorithm;

[0030] An average slope difference calculation unit for calculating the average slope difference of the accumulation sequence to obtain the forward difference of the accumulation sequence;

[0031] A start-stop information acquisition unit for determining the start-stop information of the motor-type load according to the forward difference of the accumulation sequence.

[0032] Further, it further includes: a smoothing processing unit for, after acquiring the active power sequence, selecting a fixed window, and bringing the active power sequence in the fixed window into a Gaussian function for smoothing processing; the formula is as follows:

[0033]

[0034] where f k is the signal after smoothing processing, x is the active power sequence, σ is the variance in a window, and μ is the mean value of the active power sequence x with each multiple of points as a window.

[0035] In a third aspect, the present invention provides a motor-type load event extraction device based on sliding mean optimization, including a processor and a storage medium;

[0036] The storage medium is used for storing instructions;

[0037] The processor is used for operating according to the instructions to execute the steps of the method according to any one of the foregoing.

[0038] In a fourth aspect, the present invention provides a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, the steps of the method according to any one of the foregoing are implemented.

[0039] Compared with the prior art, the beneficial effects achieved by the present invention are as follows:

[0040] 1. The present invention is a non-invasive method. By only collecting data at the user's main meter, the monitoring effect can be achieved. It uses fewer devices, has low costs, and is easy to implement, providing an important basis for government supervision.

[0041] 2. The present invention calculates the mean value of data in different windows by using a sliding window, compresses the data dimension, is insensitive to small power fluctuations, and can effectively reduce the interference during the operation of electrical appliances.

[0042] 3. Under the condition of large power fluctuations, by judging whether to perform cumulative sum calculation based on the change in the mean value calculated in different windows, the statistical test value of the forward algorithm can be effectively retained, and the information calculated in the previous section is not lost.

[0043] 4. By using the evaluation algorithm of the slope change difference, the start-stop information of electrical appliances can be accurately and effectively obtained. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 is the overall flowchart of the present invention;

[0045] Figure 2 is the flowchart of the transient event detection algorithm for sliding window mean analysis provided by the embodiment of the present invention;

[0046] Figure 3 is the active power waveform diagram provided by the embodiment of the present invention;

[0047] Figure 4 is the output result after waveform algorithm calculation provided by the embodiment of the present invention;

[0048] Figure 5 is the start-stop result calculated by the slope difference change point provided by the embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0049] The present invention will be further described below with reference to the drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention and should not be used to limit the protection scope of the present invention.

[0050] Embodiment 1

[0051] This embodiment introduces a method for extracting motor load events based on sliding mean optimization, including:

[0052] Obtain the active power sequence with a set sampling frequency at the industrial and commercial main meter to be detected that has been pre-collected;

[0053] Divide the active power sequence into multiple data windows by using a sliding window with a set size, and calculate the mean value of the active power sequence in each data window respectively;

[0054] Compare the average active power sequence values of each data window with those of the adjacent previous data window. If the average active power sequence value of the current data window is greater than that of the previous data window, obtain the cumulative sequence through the cumulative sum algorithm;

[0055] Calculate the average slope difference of the cumulative sequence to obtain the forward difference of the cumulative sequence;

[0056] Determine the start-stop information of motor loads based on the forward difference of the cumulative sequence.

[0057] For the method for extracting motor load events optimized based on sliding mean provided in this embodiment, an event monitoring is performed on a stirring device in a certain chemical industry, and the active power waveform is as Figure 3 shown; the specific application process involves the following steps:

[0058] Step 1: Install a collection device at the general meter of industrial and commercial users to be detected, and obtain the active power sequence x with one point every 0.1 s.

[0059] Step 2: Preprocess the collected active power sequence x by smoothing it using the Gaussian function to eliminate the random noise brought by sampling.

[0060] Step 2.1: According to the smoothing algorithm of the Gaussian function described in Step 2, select a fixed window n w ;

[0061] Step 2.2: Substitute the active power sequence values x in the window into the Gaussian function for smoothing processing:

[0062]

[0063] where σ = 1.5, and μ is the mean of the sequence x with a window of every 10 points;

[0064] Step 3: This method uses the change of the mean values of the front and back sliding windows to extract transient change events. The flow chart is as Figure 2 described, and the output result after Step 3 in Embodiment 1 is as Figure 4 shown:

[0065] Step 3.1: According to the event detection algorithm described in Step 3, determine the size m of the sliding window;

[0066] Step 3.2: Calculate the mean value K m for each sliding window sequence, and the specific formula is as follows:

[0067]

[0068] K n is the mean value of the active power sequence x with subscripts from n to n + m - 1;

[0069] Among them, the sliding window size m = 10;

[0070] Step 3.3: Analyze according to the mean values before and after the sliding window. When the current mean value is higher than the mean value of the previous window, perform accumulation to obtain the sequence Z;

[0071] Z n = max(0, Z n-1 + x n -(K n - K n-1 ))

[0072] Z 0 = 0

[0073] Step 4: Perform an average slope difference analysis on the sequence Z obtained in Step 3 to obtain P d And make a judgment, and finally obtain the start-stop information of the electrical appliance, as Figure 5 shown.

[0074]

[0075] P d = P r - P r-1

[0076] Among them: P r is the mean value of the sequence Z from subscript n to n + m - 1;

[0077] P d is the forward difference of the sequence P with subscript r;

[0078] Among them, the sliding window m takes 10;

[0079] When P d ∈(n, +∞), a change point occurs, otherwise no change point occurs, where n changes with different devices. In this example 1, n = 1500.

[0080] Example 2

[0081] This embodiment provides a motor load event extraction device based on sliding mean optimization, including:

[0082] An acquisition unit for acquiring the active power sequence with a set sampling frequency at the industrial and commercial main meter to be detected collected in advance;

[0083] A mean value calculation unit for dividing the active power sequence into multiple data windows by using a sliding window of a set size and respectively calculating the mean values of the active power sequences in each data window;

[0084] A judgment unit, configured to compare the average value of the active power sequence of each data window with that of the adjacent previous data window. If the average value of the active power sequence of the current data window is greater than that of the previous data window, an accumulation sequence is obtained through the accumulation sum algorithm;

[0085] An average slope difference calculation unit, configured to calculate the average slope difference of the accumulation sequence to obtain the forward difference of the accumulation sequence;

[0086] A start-stop information acquisition unit, configured to determine the start-stop information of motor loads according to the forward difference of the accumulation sequence.

[0087] A smoothing processing unit, configured to select a fixed window after obtaining the active power sequence, and substitute the active power sequence in the fixed window into a Gaussian function for smoothing processing; the formula is as follows:

[0088]

[0089] where f k is the signal after smoothing processing, x is the active power sequence, σ is the variance in a window, and μ is the mean value of the active power sequence x with each multiple of points as a window.

[0090] Embodiment 3

[0091] This embodiment provides a motor load event extraction device based on sliding mean optimization, including a processor and a storage medium;

[0092] The storage medium is used to store instructions;

[0093] The processor is configured to operate according to the instructions to execute the steps of the method according to any one of Embodiment 1.

[0094] Embodiment 4

[0095] This embodiment provides a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, the steps of the method according to any one of Embodiment 1 are implemented.

[0096] The above are only the preferred embodiments of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the technical principle of the present invention, several improvements and deformations can still be made, and these improvements and deformations should also be regarded as the protection scope of the present invention.

Claims

1. A method for extracting motor - type load events optimized based on moving average, characterized in that, it includes: Obtain the active power sequence with a set sampling frequency at the general industrial and commercial meter to be detected that has been pre - collected; Use a sliding window with a set size to divide the active power sequence into multiple data windows, and calculate the mean value of the active power sequence in each data window respectively; Compare the mean value of the active power sequence of each data window with that of the adjacent previous data window. If the mean value of the active power sequence of the current data window is larger than that of the previous data window, obtain an accumulation sequence through the accumulation sum algorithm; Calculate the average slope difference of the accumulation sequence to obtain the forward difference of the accumulation sequence; Determine the start - stop information of the motor - type load according to the forward difference of the accumulation sequence.

2. The method for extracting motor - type load events optimized based on moving average according to claim 1, characterized in that, it further includes: After obtaining the active power sequence, select a fixed window, and bring the active power sequence in the fixed window into a Gaussian function for smoothing processing. The formula is as follows: where f k is the signal after smoothing, x is the active power sequence, σ is the variance in a window, and μ is the mean of the active power sequence x with each multiple of points as the window.

3. The method for extracting motor - type load events optimized based on moving average according to claim 2, characterized in that: The step of using a sliding window with a set size to divide the active power sequence into multiple data windows and calculating the mean value of the active power sequence in each data window respectively, the formula is as follows: Among them, K n is the average value of the numerical values with subscripts from n to n + m - 1 in the active sequence x; m is the window size.

4. The method for extracting motor - type load events optimized based on moving average according to claim 3, characterized in that: Compare the average active power sequence values of each data window with those of the adjacent previous data window. If the average active power sequence value of the current data window is greater than that of the previous data window, then obtain the cumulative sequence Z through the cumulative sum algorithm n , and the formula is as follows: Z n = max(0, Z n-1 + x n -(K n - K n-1 ))。 5. The method for extracting motor - type load events optimized based on moving average according to claim 4, characterized in that: The step of calculating the average slope difference of the accumulation sequence to obtain the forward difference of the accumulation sequence, the formula is as follows: P d = P r -P r-1 where: P r is the mean value of sequence Z from subscript n to n + m - 1, and P d is the forward difference of sequence P at subscript r.

6. The method for extracting motor - type load events optimized based on moving average according to claim 5, characterized in that: The step of determining the start - stop information of the motor - type load according to the forward difference of the accumulation sequence includes: Compare the value of P d with a pre-set threshold. When the value exceeds a certain threshold, it is determined that an event occurs at this time, and the start-stop information of the motor load is obtained.

7. An apparatus for extracting motor - type load events optimized based on moving average, characterized in that, it includes: An acquisition unit for obtaining the active power sequence with a set sampling frequency at the general industrial and commercial meter to be detected that has been pre - collected; A mean value calculation unit for using a sliding window with a set size to divide the active power sequence into multiple data windows and calculating the mean value of the active power sequence in each data window respectively; A judgment unit for comparing the mean value of the active power sequence of each data window with that of the adjacent previous data window. If the mean value of the active power sequence of the current data window is larger than that of the previous data window, obtain an accumulation sequence through the accumulation sum algorithm; An average slope difference calculation unit for calculating the average slope difference of the accumulation sequence to obtain the forward difference of the accumulation sequence; A start - stop information acquisition unit for determining the start - stop information of the motor - type load according to the forward difference of the accumulation sequence.

8. The apparatus for extracting motor - type load events optimized based on moving average according to claim 7, characterized in that, it further includes: A smoothing processing unit for, after obtaining the active power sequence, selecting a fixed window and bringing the active power sequence in the fixed window into a Gaussian function for smoothing processing; the formula is as follows: where f k is the signal after smoothing, x is the active power sequence, σ is the variance in a window, and μ is the mean of the active power sequence x with each multiple of points as a window.

9. An apparatus for extracting motor - type load events optimized based on moving average, characterized in that: Comprising a processor and a storage medium; The storage medium is used for storing instructions; The processor is used for operating according to the instructions to execute the steps of the method according to any one of claims 1 to 6.

10. A computer-readable storage medium, on which a computer program is stored, characterized in that: When the program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images