A contact network operation and maintenance method and device based on time-space distribution technology

Abstract

The present invention discloses a contact network operation and maintenance method and device based on time-space distribution technology, the method comprising obtaining historical data of contact network bird nests and historical data of causal factors of contact network bird nests; constructing a contact network bird nest time distribution prediction model based on the acquired historical data, and using the time distribution prediction model to predict the contact network bird nest time distribution; training a contact network bird nest causal factor weight prediction model based on the acquired historical data, and using the contact network bird nest causal factor weight prediction model to predict the contact network bird nest spatial distribution; according to the contact network bird nest time distribution prediction result and the contact network bird nest spatial distribution prediction result, determining and outputting the contact network fault prevention and maintenance strategy. The present invention accurately predicts the contact network bird nest distribution from the two dimensions of time and space, respectively, so as to provide more comprehensive and reliable data support for contact network fault prediction and maintenance.

Description

Technical Field

[0001] The present invention belongs to the field of overhead line fault prevention and data mining technology, and specifically relates to an overhead line operation and maintenance method and device based on time-space distribution technology. Background Art

[0002] There are many factors that lead to overhead line failures. Currently, two factors are considered more: weather factors and human damage factors. Therefore, these two factors are also considered in the process of overhead line fault diagnosis and prediction.

[0003] Bird activities can easily cause problems such as short circuits, tripping, and jamming and failure of mechanical compensation devices on the overhead line, which can interfere with the transportation of electrified railways and seriously affect the power supply safety of electrified railways. The main form of bird damage to the overhead line is that birds build bird nests on the overhead line system. In order to reduce tripping and equipment failures caused by bird nests, the manpower and material resources invested by each power supply section have increased year by year.

[0004] At present, the existing contact network fault prediction technology does not take into account the impact of bird activities on contact network faults, which leads to reduced reliability and accuracy of contact network fault prediction and fails to provide more comprehensive and reliable data support and technical support for relevant operation and maintenance departments. Summary of the invention

[0005] In view of the above problems, the present invention provides a contact network operation and maintenance method based on time-space distribution technology. The present invention accurately predicts the distribution of bird nests in the contact network from the two dimensions of time and space, thereby providing more comprehensive and reliable data support for contact network fault prediction and maintenance.

[0006] The present invention is achieved through the following technical solutions:

[0007] A contact network operation and maintenance method based on time-space distribution technology, comprising:

[0008] Obtain historical data on bird nests on the contact network and historical data on the causes of bird nests on the contact network;

[0009] Based on the acquired historical data, a catenary bird nest time distribution prediction model is constructed, and the catenary bird nest time distribution prediction is performed using the time distribution prediction model;

[0010] Based on the acquired historical data, a weighted prediction model of the causal factors of bird nests on the contact network is trained, and the weighted prediction model of the causal factors of bird nests on the contact network is used to predict the spatial distribution of bird nests on the contact network;

[0011] According to the prediction results of the temporal distribution of bird nests in the contact network and the spatial distribution of bird nests in the contact network, the contact network fault prevention and maintenance strategy is determined and output.

[0012] Preferably, the process of constructing the contact network bird nest time distribution prediction model of the present invention specifically includes:

[0013] Obtain the time series of bird nest occurrence on the contact network;

[0014] Performing zero mean processing on the time series;

[0015] Calculating the autocorrelation coefficient and partial autocorrelation coefficient of the time series;

[0016] Based on the time series and its autocorrelation coefficient and partial autocorrelation coefficient, a SARIMA model is constructed;

[0017] The parameters of the catenary bird nest time distribution prediction model are determined, thereby obtaining the catenary bird nest time distribution prediction model.

[0018] Preferably, the present invention uses the contact network bird nest time distribution prediction model to predict the contact network bird nest time distribution, including:

[0019] Through the contact network bird nest time series data, the contact network bird nest time distribution prediction model is used to solve the regression coefficient, so as to predict the number of contact network bird nests in the future period.

[0020] Preferably, the present invention further comprises, after the step of obtaining the time series of occurrence of bird nests in the contact network and before the step of performing zero mean processing on the time series:

[0021] It is determined whether the time series is stationary. If so, the time series is subjected to zero mean processing. Otherwise, the time series is subjected to d times of difference processing and then subjected to zero mean processing.

[0022] Preferably, the parameters of the catenary bird nest time distribution prediction model of the present invention include:

[0023] Estimating parameters in the SARIMA model;

[0024] Using AIC to test the parameters, and obtaining AIC values ​​in the SARIMA model under different parameter groups;

[0025] The parameter group corresponding to the minimum AIC value is obtained as the parameters of the catenary bird nest time distribution prediction model.

[0026] Preferably, the present invention uses AIC to test the parameters, including:

[0027] If the test is passed, the parameters of the catenary bird nest time distribution prediction model are determined;

[0028] If the test fails, the process returns to the step of "building a SARIMA model based on the time series and its autocorrelation coefficient and partial autocorrelation coefficient".

[0029] Preferably, the catenary bird nest causal factor weight prediction model training process of the present invention specifically includes:

[0030] Obtain historical data on the number of bird nests and causal factors on the contact network and perform normalization processing to obtain a training data set;

[0031] Initialize the BP neural network model structure and variables, and train the initialized BP neural network model according to the training data set, so as to obtain a weight prediction model of the causal factors of bird nests in the contact network.

[0032] Preferably, the present invention uses the weight prediction model of the causal factors of catenary bird nests to predict the spatial distribution of catenary bird nests, including:

[0033] Using the catenary bird nest causal factor weight prediction model, the causal factor measurement data of the anchor section is obtained, so as to obtain the weight vector corresponding to each causal factor;

[0034] According to the weight vectors corresponding to the causal factors, the spatial distribution of bird nests in the contact network is predicted.

[0035] Preferably, the present invention predicts the spatial distribution of bird nests on the contact network according to the weight vectors corresponding to the causal factors, including:

[0036] Calculate the bird nest risk of all anchor sections along the entire line;

[0037] Based on the bird nest risk level of all anchor sections along the line, the high-incidence areas of the anchor sections are determined.

[0038] Preferably, the present invention trains the initialized BP neural network model according to the training data set to obtain a weight prediction model of the causal factors of bird nests in the contact network, including:

[0039] Randomly select the normalized causal factor data of the kth anchor segment and the corresponding expected output;

[0040] Calculate the input and output of each neuron in the hidden layer;

[0041] Update the weight values ​​of the causal factors through back propagation and calculate the global error;

[0042] Determining whether the weight prediction model of the catenary bird nest causative factors has converged;

[0043] If converged, stop training; otherwise, select the next training sample and the corresponding expected output, and continue the next round of learning until the weight prediction model of the causal factors of bird nests in the contact network converges.

[0044] Preferably, the present invention determines and outputs a contact network fault prevention and maintenance strategy based on the contact network bird nest time distribution prediction result and the contact network bird nest space distribution prediction result, including:

[0045] Normalize the bird nest risk of each anchor section along the entire line;

[0046] Calculate the expected number of bird nests per anchor segment in each month;

[0047] According to the estimated number of bird nests in each anchor section in each month combined with the temporal and spatial distribution of the bird nests, the high risk of bird damage to the contact network is determined and corresponding prevention strategies are formulated.

[0048] In a second aspect, the present invention proposes a contact network operation and maintenance device based on time-space distribution technology, including a data acquisition module, a time distribution prediction module, a space distribution prediction module and a prevention and maintenance module;

[0049] Wherein, the data acquisition module is used to acquire the historical data of bird nests on the contact network and the historical data of the causal factors of bird nests on the contact network;

[0050] The time distribution prediction module constructs a contact network bird nest time distribution prediction model based on the acquired historical data, and uses the time distribution prediction model to predict the contact network bird nest time distribution;

[0051] The spatial distribution prediction module trains a weighted prediction model of the causal factors of bird nests on the contact network based on the acquired historical data, and uses the weighted prediction model of the causal factors of bird nests on the contact network to predict the spatial distribution of bird nests on the contact network;

[0052] The prevention and maintenance module determines and outputs a contact network fault prevention and maintenance strategy according to the contact network bird nest time distribution prediction result and the contact network bird nest space distribution prediction result.

[0053] The present invention has the following advantages and beneficial effects:

[0054] 1. The present invention fully considers the internal and external factors that may affect the nesting of birds on the contact network, applies the BP neural network to study the spatial distribution characteristics of bird nests on the contact network, and uses the SARIMA model to study the temporal distribution laws of bird nests on the contact network, thereby realizing accurate prediction of the distribution of bird nests on the contact network from the two dimensions of time and space, respectively, and providing more comprehensive and effective data support for the prevention of contact network faults.

[0055] 2. The model of the present invention has a simple structure, easy functions, and a time scale that is accurate to the month and a spatial scale that is as fine as the anchor section, and has strong practical operability. Compared with the existing maintenance methods, it saves manpower and material resources, reduces the risk of contacting bird nests with less resource consumption, and can significantly improve the maintenance efficiency of railway operating departments. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] The drawings described herein are used to provide a further understanding of the embodiments of the present invention, constitute a part of this application, and do not constitute a limitation of the embodiments of the present invention. In the drawings:

[0057] Figure 1 The figure is a schematic diagram of a method flow of an embodiment of the present invention.

[0058] Figure 2 Schematic diagram of the SARIMA training prediction process according to an embodiment of the present invention.

[0059] Figure 3 Schematic diagram of the spatial distribution prediction process of bird nests in the contact network according to an embodiment of the present invention.

[0060] Figure 4 Schematic diagram of a computer device according to an embodiment of the present invention.

[0061] Figure 5 4 is a block diagram of the device principle of an embodiment of the present invention. DETAILED DESCRIPTION

[0062] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with embodiments and drawings. The exemplary embodiments of the present invention and their description are only used to explain the present invention and are not intended to limit the present invention.

[0063] Example 1

[0064] This embodiment proposes a contact network operation and maintenance method based on time-space distribution technology. Figure 1 As shown, the method of this embodiment includes:

[0065] Step S1, obtaining historical data of bird nests on the contact network and historical data of factors causing bird nests on the contact network.

[0066] Step S2, based on the acquired historical data, construct a time distribution prediction model for bird nests in the contact network, and use the time distribution prediction model to predict the time distribution of bird nests in the contact network.

[0067] Step S3: Based on the acquired historical data, a weighted prediction model of the causal factors of bird nests in the contact network is trained, and the weighted prediction model of the causal factors of bird nests in the contact network is used to predict the spatial distribution of bird nests in the contact network.

[0068] Step S4, determining and outputting the contact network fault prevention and maintenance strategy according to the contact network bird nest time distribution prediction result and the contact network bird nest space distribution prediction result.

[0069] The occurrence of bird nests in the contact network is continuous. By using historical bird nest data for statistical analysis, we can get the occurrence pattern of bird nest defects. The number of defects of bird nests in the contact network of the entire line each month is recorded as X(t). Since the contact network system is erected in the open air, the occurrence rate of bird nests is affected by random factors in addition to seasonal factors such as the external environment. On the other hand, the contact network components will age and wear out over time, and the occurrence rate of bird nests will show a slow upward trend in the long run.

[0070] Therefore, this embodiment adopts the seasonal autoregressive moving average model (SARIMA) to predict the number of bird nest defects in the contact network in the future based on the number of bird nest defects in the contact network in historical months (i.e., the number of bird nests in the contact network time series). Figure 2 As shown, the process of constructing the contact network bird nest time distribution prediction model in step 2 of this embodiment specifically includes:

[0071] Step S21, obtaining the time series of occurrence of bird nests in the contact network.

[0072] Step S22, determine whether the time series is stationary, if so, proceed to the subsequent steps, otherwise, perform d times difference processing on the time series.

[0073] Step S23, performing zero mean processing on the time series.

[0074] Step S24, calculating the time series autocorrelation coefficient (ACF) and partial autocorrelation coefficient (PACF).

[0075] Step S25, constructing a SARIMA model based on the time series and its autocorrelation coefficient and partial autocorrelation coefficient.

[0076] The autoregressive sliding model (ARIMA) can be expressed as:

[0077]

[0078] Among them, p is the autoregressive order, q is the sliding order, and x t is the number of bird nests on the contact network in month t, is the coefficient corresponding to the number of bird nests in the contact network in month ti, ε t is the disturbance term of month t, θ i is the coefficient of the disturbance term in period i. Before applying the above model, a d-order difference is required to make the time series stable. Then the ARIMA model can be expressed as:

[0079]

[0080] Among them, φ p (L) is the autoregressive coefficient polynomial, θ q (L) is the sliding average polynomial, It means to perform d-order difference.

[0081] The SARIMA model adds seasonal factors to ARIMA. An ARIMA model is fitted to the seasonal cyclical factors, and then a general ARIMA model is established by analogy using a seasonal time series with a period of s after difference. The comprehensive form is:

[0082]

[0083] Among them, Φ P (L s ) represents the P-order seasonal autoregressive coefficient polynomial, Θ Q (L s ) is the Q-order seasonal moving average, u t is the time series with seasonal factors removed. Substituting equation (2) into u in (3) t , we can get the complete SARIMA model in the following form:

[0084]

[0085] Among them, D represents the operation on seasonal terms, d represents the operation on non-seasonal terms, and the parameters p, q, P, Q are all determined by the values ​​of the autocorrelation coefficient (ACF) and the partial correlation coefficient (PACF) of the time series, that is, the values ​​of the orders p, q, P, Q are determined by visually identifying the truncation of the autocorrelation coefficient and the partial correlation coefficient.

[0086] Step S26, estimating parameters in the model. In this embodiment, the parameters in the model are obtained by iterating existing data, and existing iteration methods can be used, which will not be described in detail here.

[0087] Step S27, perform AIC test, if the test passes, then obtain the contact network bird nest time prediction model, otherwise return to step S25 to continue training.

[0088] This embodiment uses the AIC criterion to verify the AIC value of the model under different parameter groups, and takes the parameter group corresponding to the minimum AIC value as the final model parameter, thereby obtaining the overhead line bird nest time prediction model. Among them, the AIC detection standard is as follows:

[0089] Akaike information criterion was used to assess complexity and goodness of fit

[0090] AIC=2k-2ln L (5)

[0091] Where k is the model parameter and L is the maximum likelihood function.

[0092] The monthly statistics of the bird nest defects in the contact network have the characteristic of a "year" cycle, so the seasonal cycle s = 12. The existing contact network bird nest time series data is brought in, the regression coefficient is solved using the least squares method, and the number of contact network bird nests in the next two months is predicted.

[0093] The risk of bird nesting on the contact network is closely related to the contact network structure, weather, ecological environment, patrol work intensity and other causal factors. The causal factors and value types of bird nests on the contact network are shown in Table 1:

[0094] Table 1 Causes of bird nests in contact lines

[0095] Serial number Causative factors Value Type Remark 1 temperature float Unit: Celsius. 2 Relative humidity float unit:% 3 Water Source bool Are there any water sources such as lakes or rivers within 1km of the straight line? 4 Number of jumper shoulders int Number of triangles formed by the jumper shoulder and the line in the anchor section 5 Number of ratchet compensation devices int Number of ratchet compensation devices in the anchor section 6 Number of lattice steel columns int Number of lattice steel columns in the anchor section 7 Number of bird repellers int Number of bird repellents in anchor section 8 Planned number of inspections int Number of patrol operations covering this anchor section within the next month

[0096] This embodiment uses the above-mentioned causal factors as the input of the neural network, learns the influence weight of each causal factor through the BP neural network, and then calculates the risk of bird nests in each anchor section of the line in the future. Figure 3 As shown, the training of this embodiment obtains a weighted prediction model of the causal factors of bird nests in the contact network, and uses the weighted prediction model of the causal factors of bird nests in the contact network to predict the spatial distribution of bird nests in the contact network, specifically including:

[0097] Step S31, obtaining historical data on the number of bird nests on the contact network and causal factors and normalizing them to obtain a training data set.

[0098] In this embodiment, for the i-th type of causal factors in the historical data, Q(t) is the t-th sample data, T i (t) is the t-th training data, which is processed by min-max normalization:

[0099]

[0100] Step S32, initializing the BP neural network model structure and variables.

[0101] In this embodiment, since there are 8 types of factors causing bird nests in the contact network, the input layer and the output layer both have 8 neurons, and the hidden layer has p neurons. Then:

[0102] Hidden layer input vector: h i =(h i1 ,h i2 ,h i3 ...,h ip )

[0103] Hidden layer output vector: h o =(h o1 ,h o2,h o3 ...,h op )

[0104] Output layer input vector: y i =(y i1 ,y i2 ,y i3 ...,y i8 )

[0105] Output layer output vector: y o =(y o1 ,y o2 ,y o3 ...,y o8 )

[0106] Expected output vector: d = (d1, d2, d3..., d8)

[0107] Activation function: f = sigmoid(·)

[0108] The connection weight w between the input layer and the hidden layer ih , the connection weight w between the hidden layer and the output layer ho , hidden layer neuron threshold b h , the output layer neuron threshold b o , the number of historical anchor segment data k = 1, 2, 3...m. Assign a random number in the interval (-1, 1) to each connection weight, set the error threshold ε and the maximum number of learning times M.

[0109] Step S33, using the training data set to train the initialized BP neural network model, thereby obtaining a weight prediction model for the causal factors of bird nests in the contact network.

[0110] In this embodiment, the model training process includes:

[0111] A. Randomly select the normalized causal factor data of the kth anchor segment and the corresponding expected output;

[0112] B. Calculate the input and output of each neuron in the hidden layer;

[0113] C. Update the weight values ​​of the causal factors through back propagation and calculate the global error;

[0114] D. Determine whether the model has converged;

[0115] If convergence occurs, stop training;

[0116] Otherwise, select the next training sample and the corresponding expected output and continue the next round of learning until the model converges.

[0117] Preferably, the model training process of this embodiment is as follows:

[0118] (1) Randomly select the normalized causal factor data of the kth anchor segment and the corresponding expected output:

[0119] x=(T1(k),T2(k),...T8(k))

[0120] d=(d1(k),d2(k),...d8(k))

[0121] (2) Calculate the input and output of each neuron in the hidden layer:

[0122] h i =w ih ·x+b h (7)

[0123] h oi =f(h i (i)),i=1,2,3,...,8 (8)

[0124] y i =w ho ·h i +b o (9)

[0125] y oi =f(y i (i)),i=1,2,3,...,8 (10)

[0126] Error function:

[0127] (3) Update weight values ​​through back propagation:

[0128]

[0129]

[0130] w ho N and w ho N It represents the weight during the Nth training, and μ takes the value of 0.05.

[0131] (4) Calculate the global error:

[0132]

[0133] (5) Determine whether the model converges

[0134] Determine whether the BP neural network error meets the requirements. When the error is less than the preset accuracy ε or the number of learning times reaches M, stop training; otherwise, select the next training sample and the corresponding expected output to enter the next round of learning.

[0135] Step S34, inputting the measured values ​​of the causal factors of the anchor section into the weight prediction model of the causal factors of the contact network bird nests to obtain the weight vector corresponding to each causal factor.

[0136] Step S35, predicting the spatial distribution of bird nests on the contact network according to the weight vectors corresponding to the various causal factors.

[0137] In this embodiment, It is the weight vector corresponding to each cause, that is, the learning result of BP neural network.

[0138] According to the weight vectors corresponding to various causal factors, the spatial distribution of bird nests in the contact network is predicted, including:

[0139] Step S351, calculating the bird nest risk of all anchor sections of the line.

[0140] The calculation formula for the bird nest risk of a specific anchor section is as follows:

[0141]

[0142] Among them, R is the bird nest risk of the anchor section, is the weight vector of the causal factors, is the causal factor vector.

[0143] Step S352, determining the high-incidence areas of the anchor sections according to the calculated bird nest risk levels of all anchor sections of the line.

[0144] Preferably, in this embodiment, the top five or ten anchor segments with the highest bird nest risk can be selected as bird nest high-incidence areas.

[0145] Step 4 of this embodiment specifically includes:

[0146] This embodiment formulates a targeted prevention and maintenance plan based on the prediction results of the time distribution of bird nests in the contact network and the spatial distribution of the contact network, so as to realize the transition of the contact network maintenance from the existing situation-based maintenance to condition-based maintenance. The specific implementation method is as follows:

[0147] Step S41, normalizing the bird nest risk of each anchor section of the entire line.

[0148] According to the spatial distribution prediction model of bird nests in the overhead line, the risk list of all anchor sections of the entire line in the next two months is calculated. R1 = [R 11 ,R 12 ,...,R 1n ],R2=[R 21 ,R 22 ,...,R 2n ], the risk of each anchor segment in the above list is normalized as follows:

[0149]

[0150] Among them, R i Represents the risk list for the next i months, R ij represents the risk of the jth anchor segment in the future i months, is the normalized risk of the jth anchor section in the next i month. Then the normalized risk list of all anchor sections of the entire line in the next two months is

[0151] Step S42, calculating the estimated number of bird nests in each anchor segment in each month.

[0152] The bird nest time distribution prediction model shows that the total number of bird nests on the entire line in the next two months will be C1 and C2 respectively. The expected number of bird nests in the jth anchor section in the next i month is calculated by the following formula:

[0153] N ij =C i ·R ij (17)

[0154] The number of bird nests in each anchor section in the next two months is N1 = [N 11 ,N 12 ,...,N 1n ] and N2=[N 21 ,N 22 ,...,N 2n ].

[0155] Step S43, formulate a prevention and maintenance plan.

[0156] Specifically, based on the expected number of bird nests in each anchor section each month, combined with the temporal and spatial distribution of bird nests, the high risk of bird damage to the contact network is determined, and corresponding prevention strategies are formulated.

[0157] Preferably, this embodiment marks the top five anchor section areas and time marks in N1 and N2 as high risk of bird damage to the contact network through joint statistics, and formulates corresponding maintenance strategies. For example, the inspection frequency of the above-mentioned anchor sections at the corresponding time is ensured to be targeted; at the same time, the inspection frequency of anchor sections where the number of bird nests is expected to be less than 2 is reduced to save manpower and material resources.

[0158] This embodiment also provides a computer device for executing the above method of this embodiment.

[0159] Specific as Figure 4As shown, the computer device includes a processor, an internal memory and a system bus; various device components including the internal memory and the processor are connected to the system bus. The processor is a piece of hardware used to execute computer program instructions through basic arithmetic and logical operations in the computer system. The internal memory is a physical device used to temporarily or permanently store computer programs or data (for example, program status information). The system bus can be any of the following types of bus structures, including a memory bus or a storage controller, a peripheral bus, and a local bus. The processor and the internal memory can communicate data through the system bus. The internal memory includes a read-only memory (ROM) or a flash memory (not shown in the figure), and a random access memory (RAM), and RAM usually refers to the main memory loaded with the operating system and computer programs.

[0160] Computer devices generally include an external storage device. The external storage device can be selected from a variety of computer-readable media, which refers to any available media that can be accessed by the computer device, including both removable and fixed media. For example, computer-readable media include, but are not limited to, flash memory (micro SD card), CD-ROM, digital versatile disk (DVD) or other optical disk storage, magnetic tape cassettes, magnetic tapes, magnetic disk storage or other magnetic storage devices, or any other medium that can be used to store the desired information and can be accessed by the computer device.

[0161] Computer devices can be logically connected to one or more network terminals in a network environment. Network terminals can be personal computers, servers, routers, smart phones, tablets, or other public network nodes. Computer devices are connected to network terminals through network interfaces (local area network LAN interfaces). Local area network (LAN) refers to a network of computers interconnected within a limited area, such as a home, school, computer laboratory, or office building using network media. WiFi and twisted pair cabling Ethernet are the two most commonly used technologies for building LANs.

[0162] It should be noted that other computer systems including more or fewer subsystems than a computer device may also be suitable for the invention.

[0163] As described in detail above, the computer device applicable to this embodiment can perform the specified operations of the contact network operation and maintenance method based on time-space distribution technology. The computer device performs these operations in the form of software instructions running in a computer-readable medium through a processor. These software instructions can be read into the memory from a storage device or from another device through a local area network interface. The software instructions stored in the memory enable the processor to execute the above-mentioned group member information processing method. In addition, the present invention can also be implemented by hardware circuits or hardware circuits combined with software instructions. Therefore, the implementation of this embodiment is not limited to any specific combination of hardware circuits and software.

[0164] Example 2

[0165] This embodiment proposes a contact network operation and maintenance device based on time-space distribution technology. Figure 5 As shown, it includes a data acquisition module, a time distribution prediction module, a space distribution prediction module, and a prevention and maintenance module.

[0166] Among them, the data acquisition module is used to obtain the historical data of bird nests on the contact network and the historical data of the causal factors of bird nests on the contact network.

[0167] The time distribution prediction module constructs a contact network bird nest time distribution prediction model based on the acquired historical data, and uses the time distribution prediction model to predict the contact network bird nest time distribution.

[0168] The spatial distribution prediction module trains a weighted prediction model of the causal factors of bird nests in the contact network based on the acquired historical data, and uses the weighted prediction model of the causal factors of bird nests in the contact network to predict the spatial distribution of bird nests in the contact network.

[0169] The prevention and maintenance module determines and outputs the contact network fault prevention and maintenance strategy based on the prediction results of the time distribution of bird nests in the contact network and the prediction results of the spatial distribution of bird nests in the contact network.

[0170] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A contact network operation and maintenance method based on time-space distribution technology, characterized in that: include: Obtain historical data on bird nests on the contact network and historical data on the causes of bird nests on the contact network; Based on the acquired historical data, a catenary bird nest time distribution prediction model is constructed, and the catenary bird nest time distribution prediction is performed using the time distribution prediction model; Based on the acquired historical data, a weighted prediction model of the causal factors of bird nests on the contact network is trained, and the weighted prediction model of the causal factors of bird nests on the contact network is used to predict the spatial distribution of bird nests on the contact network; The spatial distribution prediction of bird nests on the contact network is performed using the weight prediction model of the causal factors of bird nests on the contact network, including: Using the catenary bird nest causal factor weight prediction model, the causal factor measurement data of the anchor section is obtained, so as to obtain the weight vector corresponding to each causal factor; According to the weight vectors corresponding to the causal factors, the spatial distribution of bird nests on the contact network is predicted; the spatial distribution of bird nests on the contact network is predicted according to the weight vectors corresponding to the causal factors, including: Calculate the bird nest risk of all anchor sections along the entire line; the calculation formula for the bird nest risk of anchor sections is: Among them, R is the bird nest risk of the anchor section, is the weight vector of the causal factors, is the causal factor vector; According to the bird nest risk level of all anchor sections of the line, determine the high-incidence areas of the anchor sections; According to the prediction results of the time distribution of bird nests in the contact network and the prediction results of the spatial distribution of bird nests in the contact network, the fault prevention and maintenance strategies of the contact network are determined and output; According to the prediction results of the time distribution of bird nests in the contact network and the prediction results of the spatial distribution of bird nests in the contact network, the contact network fault prevention and maintenance strategy is determined and output, including: Normalize the bird nest risk of each anchor section of the entire line; specifically calculate the risk list of all anchor sections of the entire line in the next two months according to the contact network bird nest spatial distribution model, and normalize the risk of each anchor section in the risk list of the two months to obtain a normalized risk list of all anchor sections of the entire line in the next two months; Calculate the expected number of bird nests for each anchor section in each month; specifically, calculate the total number of bird nests for the entire line in the next two months based on the bird nest time distribution prediction model; calculate the number of bird nests for all anchor sections of the entire line in the next two months based on the total number of bird nests for the entire line in two months and the risk list of all anchor sections in the next two months; According to the estimated number of bird nests in each anchor section in each month, combined with the temporal and spatial distribution of bird nests, the high risk of bird damage to the contact network is determined, and corresponding prevention strategies are formulated.

2. The contact network operation and maintenance method according to claim 1, characterized in that: The construction process of the contact network bird nest time distribution prediction model specifically includes: Obtain the time series of bird nest occurrence on the contact network; Performing zero mean processing on the time series; Calculating the autocorrelation coefficient and partial autocorrelation coefficient of the time series; Based on the time series and its autocorrelation coefficient and partial autocorrelation coefficient, a SARIMA model is constructed; The parameters of the catenary bird nest time distribution prediction model are determined, thereby obtaining the catenary bird nest time distribution prediction model.

3. The contact network operation and maintenance method according to claim 2, characterized in that: The method of using the contact network bird nest time distribution prediction model to predict the contact network bird nest time distribution includes: Through the contact network bird nest time series data, the contact network bird nest time distribution prediction model is used to solve the regression coefficient, so as to predict the number of contact network bird nests in the future period.

4. The contact network operation and maintenance method according to claim 2, characterized in that: After the step of obtaining the time series of bird nest occurrence in the contact network, before the step of performing zero mean processing on the time series, the step further includes: It is determined whether the time series is stationary. If so, the time series is subjected to zero mean processing. Otherwise, the time series is subjected to d times of difference processing and then subjected to zero mean processing.

5. The contact network operation and maintenance method according to claim 2, characterized in that: The parameters of the contact network bird nest time distribution prediction model are determined, including: Estimating parameters in the SARIMA model; Using AIC to test the parameters, and obtaining AIC values ​​in the SARIMA model under different parameter groups; The parameter group corresponding to the minimum AIC value is obtained as the parameters of the catenary bird nest time distribution prediction model.

6. The contact network operation and maintenance method according to claim 5, characterized in that: The parameters tested using AIC include: If the test is passed, the parameters of the catenary bird nest time distribution prediction model are determined; If the test fails, return to the step of "building a SARIMA model based on the time series and its autocorrelation coefficient and partial autocorrelation coefficient".

7. The contact network operation and maintenance method according to claim 1, characterized in that: The training process of the weight prediction model of the factors causing bird nests in the contact network specifically includes: Obtain historical data on the number of bird nests and causal factors on the contact network and perform normalization processing to obtain a training data set; Initialize the BP neural network model structure and variables, and train the initialized BP neural network model according to the training data set, so as to obtain a weight prediction model of the causal factors of bird nests in the contact network.

8. The contact network operation and maintenance method according to claim 7, characterized in that: The initialized BP neural network model is trained according to the training data set to obtain a weight prediction model of the factors causing bird nests in the contact network, including: Randomly select the normalized causal factor data of the kth anchor segment and the corresponding expected output; Calculate the input and output of each neuron in the hidden layer; Update the weight values ​​of the causal factors through back propagation and calculate the global error; Determining whether the weight prediction model of the catenary bird nest causative factors has converged; If converged, stop training; otherwise, select the next training sample and the corresponding expected output, and continue the next round of learning until the weight prediction model of the causal factors of bird nests in the contact network converges.

9. A contact network operation and maintenance device based on time-space distribution technology, characterized in that: It includes data acquisition module, time distribution prediction module, space distribution prediction module and prevention and maintenance module; Wherein, the data acquisition module is used to acquire the historical data of bird nests on the contact network and the historical data of the causal factors of bird nests on the contact network; The time distribution prediction module constructs a contact network bird nest time distribution prediction model based on the acquired historical data, and uses the time distribution prediction model to predict the contact network bird nest time distribution; The spatial distribution prediction module trains a weighted prediction model of the causal factors of bird nests on the contact network based on the acquired historical data, and uses the weighted prediction model of the causal factors of bird nests on the contact network to predict the spatial distribution of bird nests on the contact network; using the weighted prediction model of the causal factors of bird nests on the contact network to predict the spatial distribution of bird nests on the contact network includes: Using the catenary bird nest causal factor weight prediction model, the causal factor measurement data of the anchor section is obtained, so as to obtain the weight vector corresponding to each causal factor; According to the weight vectors corresponding to the causal factors, the spatial distribution of bird nests on the contact network is predicted; the spatial distribution of bird nests on the contact network is predicted according to the weight vectors corresponding to the causal factors, including: Calculate the bird nest risk of all anchor sections along the entire line; the calculation formula for the bird nest risk of anchor sections is: Among them, R is the bird nest risk of the anchor section, is the weight vector of the causal factors, is the causal factor vector; According to the bird nest risk level of all anchor sections of the line, determine the high-incidence areas of the anchor sections; The prevention and maintenance module determines and outputs a contact network fault prevention and maintenance strategy based on the contact network bird nest time distribution prediction result and the contact network bird nest space distribution prediction result; According to the prediction results of the time distribution of bird nests in the contact network and the prediction results of the spatial distribution of bird nests in the contact network, the contact network fault prevention and maintenance strategy is determined and output, including: Normalize the bird nest risk of each anchor section of the entire line; specifically calculate the risk list of all anchor sections of the entire line in the next two months according to the contact network bird nest spatial distribution model, and normalize the risk of each anchor section in the risk list of the two months to obtain a normalized risk list of all anchor sections of the entire line in the next two months; Calculate the expected number of bird nests for each anchor section in each month; specifically, calculate the total number of bird nests for the entire line in the next two months based on the bird nest time distribution prediction model. Based on the total number of bird nests for the entire line in two months and the normalized risk list of all anchor sections in the next two months, calculate the number of bird nests for all anchor sections of the entire line in the next two months; According to the estimated number of bird nests in each anchor section in each month, combined with the temporal and spatial distribution of bird nests, the high risk of bird damage to the contact network is determined, and corresponding prevention strategies are formulated.

Images