CN108182296A - A Method for Orbit Data Processing of Orbit Inspection Instrument Based on Empirical Mode Decomposition - Google Patents

Abstract

The invention discloses a method for processing orbital data of an orbit detector based on empirical mode decomposition, which comprises the following steps: Step 1: input a set of original orbital data; step 2: use empirical mode decomposition to process the original orbital data Decompose to obtain the intrinsic mode function and trend item of each layer respectively; Step 3: Use the 3σ criterion to identify and eliminate the coarse error of the first layer intrinsic mode function IMF ₁ obtained by the decomposition; Step 4: Reconstruct to obtain and remove the gross error subsequent trajectory data. The data processing method of the invention can protect detail signals while eliminating coarse errors, and is very suitable for processing nonlinear and non-stationary signals.

Description

A Method for Orbit Data Processing of Orbit Inspection Instrument Based on Empirical Mode Decomposition

technical field

The invention belongs to the field of digital signal processing, and in particular relates to a signal filtering method based on empirical mode decomposition corresponding to intentionally suggested data collection in railway engineering safety detection.

Background technique

Rail is the basic equipment of railway transportation, and its performance is directly related to the stability and safety of the train. Because the rail is exposed to various natural environments all the year round, and with the continuous action of the train load, the geometric size of the rail will change. Constant changes occur, so line inspection is a necessary means to understand the quality status of the rails. Commonly used line detection is divided into track dynamic detection and track static detection. The so-called track static detection refers to the inspection of the track by using detection tools or equipment such as track rulers, strings and track detectors. The content of the inspection mainly includes the geometric dimensions of the track such as gauge, level (distortion), height, and direction, as well as the status of components such as rails, coupling fasteners, ballast beds, and turnouts.

The track inspection instrument is equipped with various high-precision sensors, radio communication equipment, and outdoor computers. With the help of professional software, it is used to control measurement and data storage management. The data processing speed is fast, and the collected data can be analyzed and alarmed in time for guidance. For on-site maintenance, its sensors can detect data such as gauge, height, level, direction and mileage of the rail. The original data collected by the track inspection instrument is often mixed with coarse error noise caused by environmental factors. If these noises are not properly processed, the obtained data will be inaccurate, which will seriously affect the construction personnel's understanding of the technical status and The understanding of the law of change increases the potential safety hazard.

Therefore, it is necessary to design a signal processing method to remove the coarse error noise in the original signal, improve the accuracy of the data, and the obtained track geometry data can more faithfully reflect the state of the rail.

Contents of the invention

The technical problem to be solved by the present invention is to propose a method for processing track direction data of track detectors based on empirical mode decomposition, and use empirical mode decomposition to process the track direction data collected by track detectors. Processing to remove the coarse errors contained in the signal, so as to obtain more accurate orbit data.

Technical scheme of the present invention is:

A method for processing trajectory data of an orbit inspection instrument based on empirical mode decomposition, comprising the following steps:

Step 1: Input a set of original orbital data collected by orbit detectors;

Step 2: Perform empirical mode decomposition on the original orbital data to obtain the intrinsic mode functions IMF ₁ (t) to IMF _n (t) and remainder r _n (t) of each layer respectively;

Step 3: Identify the coarse error points in the first layer intrinsic mode function IMF ₁ (t) and eliminate them to obtain the data sequence IMF ₁ ′(t);

Step 4: Reconstruct to obtain the orbit data after removing gross errors.

Described step 2 specifically comprises the following steps:

Step 2.1: Determine all the maximum and minimum points of the original orbital data sequence x(t), perform cubic spline interpolation on them respectively, and construct the upper and lower envelope x _up (t) of x(t) and x _low (t), calculate the mean value m ₁ (t) of the upper and lower envelopes = (x _up (t)+x _low (t))/2;

Step 2.2: Calculate the difference between x(t) and m ₁ (t) to obtain a new data sequence h ₁ (t): h ₁ (t)=x(t)-m ₁ (t);

Step 2.3: judging whether h ₁ (t) is an intrinsic mode function;

For a sequence to be an eigenmode function, the following two conditions need to be satisfied:

1) In the entire time range, the number of local extremum points and the number of zero-crossing points must be equal or differ by at most one; 2) The envelope (upper envelope) formed by the local maximum and the local minimum The mean value of the formed envelope (lower envelope) is zero;

If h ₁ (t) satisfies the above two conditions, set IMF ₁ = h ₁ (t), and calculate the difference r ₁ (t) between the original orbit data sequence x(t) and IMF ₁ (t): r ₁ (t)=x(t)-IMF ₁ (t);

If h ₁ (t) does not meet the above conditions, then regard h ₁ (t) as a new data sequence, repeat step 2.1 and step 2.2, and calculate the envelope average value m ₁₁ (t) and h ₁ (t) and m ₁₁ (t), and judge whether h ₁₁ (t) is an intrinsic mode function; repeat the above process until h _1k (t) _satisfies the intrinsic mode function Define the condition, let h _1k (t)=h _1(k-1) (t)-m _1k (t) be the first intrinsic mode function component IMF ₁ (t) of x(t), and find The difference r ₁ (t) between the original orbit data sequence and IMF ₁ (t), namely r ₁ (t)=x(t)-IMF ₁ (t);

In the step 2, the IMF ₁ of the first layer contains gross errors of the original data, which need to be detected and eliminated.

In the step 3, use the 3σ criterion to identify and eliminate the gross errors in the intrinsic mode function of the first layer: the 3σ criterion is that for an arbitrary data x _d , if

It is considered that x _d is a gross error point, and the data needs to be eliminated, otherwise it will be retained; where n is the number of detected data, is the mean value of the test data, σ is the standard deviation,

In the step 4, based on the IMF ₁ ′(t) of the first layer after the gross error points are removed, and the intrinsic mode functions and remainders of the other layers, the reconstruction obtained after the gross error is removed Orbit data x′(t):

In the step 1, the original trajectory data is read from the ".csv" file that saves the data collected by the orbit detector.

In the step 4, the reconstructed orbital data is output in the ".csv" file format and saved in the computer, and drawing is performed according to the data in the processed ".csv" file, which can visually display the detection results , to guide the railway engineering department to assess the technical status of the line section.

Beneficial effect:

Due to factors such as the environment, when the track detector collects data, there will be many rough error points in the data. If these data are not processed, other data calculated from these data will be inaccurate. Traditional denoising methods such as mean filtering can eliminate noise, but signals under special circumstances will introduce more feature loss and signal distortion, and the calculation speed is slow, which is not conducive to the data processing and storage of the track inspection instrument. Empirical mode decomposition is a data-driven adaptive signal decomposition method, which can decompose data into a set of intrinsic mode function (IntrinsicMode Function, IMF) components with physical meaning. The empirical mode decomposition method does not need to set Based on the basis function, it is adaptively obtained through iteration according to the signal characteristics, which is especially suitable for dealing with nonlinear and non-stationary signals. The data collected by the orbit detector is nonlinear and non-stationary signal, and the present invention can effectively remove the coarse error in the original orbit data by taking advantage of the empirical mode decomposition, and at the same time reduce the influence of the error on the data to a certain extent, It is beneficial to the processing and subsequent analysis of the data collected by the track inspection instrument.

Description of drawings

Fig. 1 is a schematic flow chart of the present invention;

Fig. 2 is the schematic flow chart of empirical mode decomposition method;

Fig. 3 is a schematic diagram of the original orbital data waveform in the present invention;

Fig. 4 is the eigenmode component and remainder figure that the original orbit data of the present invention obtains through empirical mode decomposition;

Fig. 5 is an effect diagram of the present invention using the 3σ criterion to detect and eliminate coarse errors of the first-layer intrinsic mode function IMF ₁ of the orbital raw data, wherein Fig. 5(a) is a graph of the first-layer intrinsic mode function IMF ₁ , Fig. 5 (b) is the graph of data after processing;

Fig. 6 is the filtering effect diagram of the filtering method of the present invention and the traditional mean value filter filtering method to the orbital original data, wherein Fig. 6 (a) is a filtering effect diagram of the method of the present invention, and Fig. 6 (b) is a traditional mean value filter Filter effect diagram.

Detailed ways

In order to make the object, technical solution and advantages of the present invention clearer, the present invention will be further described in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described here are only used to explain the present invention, not to limit the present invention.

Fig. 1 shows a schematic flow chart provided by the present invention, and this embodiment is realized by MATLAB tool. The method for processing track direction data of a track inspection instrument based on empirical mode decomposition disclosed by the present invention comprises the following steps:

Step 1: Input a set of original orbital data collected by track detectors; it represents the original track direction data collected by track detectors in the mileage section of the line, and use MATLAB software to read this file into the workspace, as shown in Figure 3 It is a schematic diagram of the original orbital data waveform in the present invention;

Step 2: Carry out empirical mode decomposition on the original orbital data, and obtain the intrinsic mode functions IMF ₁ (t) to IMF _n (t) and the remainder r _n (t) of each layer respectively, as shown in Figure 4 shown;

Step 3: Use the 3σ criterion to identify the coarse error points in the first-layer intrinsic mode function IMF ₁ (t) and eliminate them to obtain the data sequence IMF ₁ ′(t), the graph of which is shown in Figure 5;

Step 4: Restructure the data. Reconstruction is the inverse process of decomposition. Add IMF ₁ ′, IMF ₂ ～IMF _n and the remainder r _n to obtain the data sequence x'(t) after filtering gross errors. Save the processed data x'(t) in a file ending in ".csv" format for subsequent use.

Fig. 2 has shown the schematic flow chart of empirical mode decomposition method in the present invention; The step of carrying out empirical mode decomposition to original orbit data sequence x (t) in above-mentioned step 2 as shown in Fig. 2 is as follows:

Step 2.1: Determine all the maximum and minimum points of the original orbital data sequence x(t), perform cubic spline interpolation on them respectively, and construct the upper and lower envelope x _up (t) of x(t) and x _low (t), calculate the mean value m ₁ (t) of the upper and lower envelopes = (x _up (t)+x _low (t))/2;

Step 2.2: Calculate the difference between x(t) and m ₁ (t) to obtain a new data sequence h ₁ (t): h ₁ (t)=x(t)-m ₁ (t);

Step 2.3: judging whether h ₁ (t) is an intrinsic mode function;

For a sequence to be an eigenmode function, the following two conditions need to be satisfied:

1) In the entire time range, the number of local extremum points and the number of zero-crossing points must be equal or differ by at most one; 2) The envelope (upper envelope) formed by the local maximum and the local minimum The mean value of the formed envelope (lower envelope) is zero;

If h ₁ (t) satisfies the above two conditions, set IMF ₁ = h ₁ (t), and calculate the difference r ₁ (t) between the original orbit data sequence x(t) and IMF ₁ (t): r ₁ (t)=x(t)-IMF ₁ (t);

If h ₁ (t) does not meet the above conditions, then regard h ₁ (t) as a new data sequence, repeat step 2.1 and step 2.2, and calculate the envelope average value m ₁₁ (t) and h ₁ (t) and m ₁₁ (t), and judge whether h ₁₁ (t) is an intrinsic mode function; repeat the above process until h _1k (t) _satisfies the intrinsic mode function Define the condition, let h _1k (t)=h _1(k-1) (t)-m _1k (t) be the first intrinsic mode function component IMF ₁ (t) of x(t), and find The difference r ₁ (t) between the original orbit data sequence and IMF ₁ (t), namely r ₁ (t)=x(t)-IMF ₁ (t);

In order to verify the effectiveness of the present invention, a comparative test is carried out between the filtering method of the present invention and the traditional mean filter filtering method, and the simulation result diagram shown in FIG. 6 is obtained. It can be seen from FIG. 6 that the use of the mean value filter for filtering processing is likely to cause signal distortion, but the method of the present invention can obviously remove the noise in the signal.

The present invention can effectively detect and eliminate gross errors in the original track direction data collected by the track detector, so as to ensure the accuracy of the data. The filtered data can be applied to the follow-up evaluation of the track state by the railway public works department, and on-site guidance for maintenance and review and acceptance work.

The above embodiments are only a part of the embodiments of the present invention. For those of ordinary skill in the art, on the premise of not paying creative work, the replacement schemes that are substantially the same as those of the present invention all belong to the protection scope of the present invention.

Claims (6)

1. a kind of rail based on empirical mode decomposition examines tradition to data processing method, which is characterized in that includes the following steps：

Step 1：The original rail of one group of rail inspection instrument acquisition is inputted to data；

Step 2：Empirical mode decomposition is carried out to data to the original rail, respectively obtains the intrinsic mode function IMF of each layer₁(t)~ IMF_n(t) and remainder r_n(t)；

Step 3：Identify first layer intrinsic mode function IMF₁(t) gross error point in is simultaneously rejected, and obtains data sequence IMF₁′(t)；

Step 4：The rail after obtaining removal gross error is reconstructed to data.

2. the rail according to claim 1 based on empirical mode decomposition examines tradition to data processing method, which is characterized in that The step 2 specifically includes following steps：

Step 2.1：It determines all maximum points and minimum point of the original rail to data sequence x (t), it is carried out three times respectively Spline interpolation constructs the envelope x up and down of x (t)_up(t) and x_low(t), the mean value m of envelope up and down is calculated₁(t)=(x_up (t)+x_low(t))/2；

Step 2.2：Calculate x (t) and m₁(t) difference between obtains a new data sequence h₁(t)：h₁(t)=x (t)-m₁ (t)；

Step 2.3：Judge h₁(t) whether it is an intrinsic mode function；

One sequence needs to meet following two conditions for intrinsic mode function：

1) in entire time range, the number of Local Extremum must equal or at most differ one with the number of zero crossing It is a；2) envelope (coenvelope line) being made of local maximum and the envelope being made of local minimum are (lower to wrap Winding thread) average value be zero；

If h₁(t) meet two above condition, enable IMF₁=h₁(t), and original rail is sought to data sequence x (t) and IMF₁(t) it Between difference r₁(t)：r₁(t)=x (t)-IMF₁(t)；

If h₁(t) conditions above is unsatisfactory for, then by h₁(t) it is considered as a new data sequence, repeats step 2.1 and step 2.2, seek its envelope average value m₁₁(t) and h₁(t) and m₁₁(t) the difference h between₁₁(t), and judge h₁₁(t) whether it is one intrinsic Mode function；Repeat the above process, until h_1k(t) meet the definition condition of intrinsic mode function, then enable h_1k(t)= h_1(k-1)(t)-m_1k(t) first intrinsic mode function component IMF for being x (t)₁(t), and ask original rail to data sequence with IMF₁(t) the difference r between₁(t), i.e. r₁(t)=x (t)-IMF₁(t)；

Step 2.4：By r₁(t) original data sequence new as one repeats step 2.1~step 2.3, obtains second sheet Levy mode function component IMF₂(t), r is enabled₂(t)=r₁(t)-IMF₂(t), iterative step 2.4 successively, until obtained eigen mode State function component IMF_n(t) no Riding Waves, swing up and down around Zero value line and when peak and valley of part is symmetrical terminates, at this time Remainder r_n(t)=r_(n-1)(t)-IMF_n(t), the empirical mode decomposition of x (t) is completed to get to the table of empirical mode decomposition (EMD) Up to formula

3. the rail according to claim 1 based on empirical mode decomposition examines tradition to data processing method, which is characterized in that In the step 3, identify the gross error in first layer intrinsic mode function using 3 σ criterion and rejected：3 σ criterion refer to For an arbitrary data x_dIf

Then think x_dIt is a gross error point, needs to reject the data, otherwise be retained；In formula, n is of detection data Number,For the average value of detection data, σ is standard deviation,

4. the rail according to claim 3 based on empirical mode decomposition examines tradition to data processing method, which is characterized in that In the step 4, based on the first layer intrinsic mode function IMF after rejecting gross error point₁The sheet of ' (t) and other each layers Mode function and remainder are levied, reconstructs the rail after obtaining removal gross error to data x ' (t)：

5. rail according to any one of claims 1 to 4 based on empirical mode decomposition examines tradition to data processing method, It is characterized in that, in the step 1, original rail is read to data from " .csv " file for preserving rail inspection instrument gathered data.

6. rail according to any one of claims 1 to 4 based on empirical mode decomposition examines tradition to data processing method, It is characterized in that, in the step 4, obtained rail will be reconstructed to data, exported with " .csv " file format and be stored in calculating In machine.