Catenary status assessment method and system based on adaptive weighted optimization

Abstract

The present invention discloses a contact network state assessment method and system based on adaptive weighted optimization; the method comprises collecting dynamic and static parameters of the contact network, calculating the subjective and objective weights of each parameter by adopting a hierarchical analysis method and an entropy weight method, and calculating the static composite weight by adopting a combined weight method; collecting historical degradation data of the dynamic and static parameters of the contact network, and obtaining the expected degradation rate of each parameter by adopting a time series neural network; calculating the real-time degradation rate of the parameter, and dynamically optimizing the static composite weight by adopting an adaptive method based on the real-time degradation rate and the expected degradation rate to obtain a dynamic composite weight optimization result; optimizing and updating the gain in the adaptive method; and performing a contact network state assessment based on the dynamic composite weight optimization result; the method and system proposed by the present invention are not only suitable for contact network state assessment, but can also be extended to other state assessment fields, and have broad application prospects.

Description

Technical Field

[0001] The present invention belongs to the field of state assessment, and in particular relates to a contact network state assessment method and system based on adaptive weighted optimization. Background Art

[0002] Currently, catenary condition assessment primarily relies on static weighting methods, such as principal component analysis, entropy weighting, and the analytic hierarchy process. While these methods can comprehensively consider expert experience and data characteristics, they still have significant limitations in practical application. Traditional methods typically employ fixed weights, which cannot effectively reflect the dynamic degradation characteristics of catenary parameters over long-term operation, resulting in deviations between assessment results and actual conditions. Furthermore, existing technologies often rely on current measurement data for assessment, lacking modeling and analysis of historical parameter degradation trends. This makes it difficult to reflect the degradation patterns of key performance indicators (such as contact force, conductor height, and hard points), thus compromising the accuracy of preventive maintenance. Furthermore, the electrical performance (such as voltage fluctuation) and mechanical properties (such as hard point amplitude) of the catenary are susceptible to environmental factors, but existing assessment systems fail to incorporate these dynamic variables into the weight optimization process, reducing the adaptability of the assessment. More critically, traditional methods rely on periodic measurement data, making it difficult to promptly capture sudden parameter degradation, resulting in delayed operational and maintenance responses and impacting the safety and reliability of railway operations. Therefore, there is an urgent need for a contact network status assessment method that can integrate multi-source data, dynamically adjust weights and have real-time optimization capabilities, while consuming low computing resources, in order to improve assessment accuracy and operation and maintenance efficiency. Summary of the Invention

[0003] In order to solve the problems existing in the background technology, the purpose of the present invention is to provide a method for evaluating the state of a contact network based on adaptive weighted optimization, which method includes the following steps:

[0004] Step (1): establishing a contact network status evaluation system, wherein the evaluation system includes an indicator level and a parameter level, wherein the indicator level includes a safety indicator X1, a smoothness indicator X2, a current collecting performance indicator X3, and an electrical performance indicator X4, and the parameter level includes a number of parameters reflecting each indicator level;

[0005] Step (2): collecting dynamic and static parameters of the contact network and processing the dynamic and static parameters of the contact network;

[0006] Step (3): Calculate the subjective and objective weights of each parameter at the parameter level using the analytic hierarchy process and entropy weight method respectively, and calculate the static composite weight of each parameter at the parameter level using the combined weight method;

[0007] Step (4): collecting historical degradation data of the dynamic and static parameters of the contact network, using a time series neural network to establish a degradation rate prediction model for each parameter of the parameter level, and obtaining the expected degradation rate of each parameter through the degradation rate prediction model; calculating the real-time degradation rate of the dynamic and static parameters of the contact network, and based on the real-time degradation rate and the expected degradation rate, using an adaptive method to dynamically optimize the static composite weights of each parameter of the parameter level in step (3) to obtain a dynamic composite weight optimization result, wherein the adaptive method includes a first adaptive gain and a second adaptive gain;

[0008] Step (5): Based on the dynamic composite weight optimization results, the state of the contact network at the data collection point is evaluated;

[0009] Step (6): constructing a cost function based on the second-order degradation rate differential and solving the cost function using the function extreme value method, and optimizing and updating the first adaptive gain and the second adaptive gain in step (4);

[0010] Repeat the above steps (4) to (6) to obtain the contact network status evaluation results at the data collection points at different times.

[0011] Furthermore, in step (2), the dynamic and static parameters of the contact network are processed, including cleaning and normalizing the parameters to remove noise and abnormal values.

[0012] Furthermore, in step (3), the subjective and objective weights of each parameter at the parameter level are calculated using the analytic hierarchy process and the entropy weight method, respectively, and the static composite weight of each parameter at the parameter level is calculated using the combined weight method, including the following steps:

[0013] Step (3.1): Use the hierarchical analysis method to calculate the parameter level reflection index The subjective weight of each parameter is used to quantify the relative importance of the parameters through expert scoring, and a judgment matrix is ​​constructed. ;

[0014] in, Indicates reflection indicators The parameter level judgment matrix, Indicates that the parameter level reflects the index No. Parameters and The relative importance of the parameters, , n Reflect indicators at the parameter level The number of parameters;

[0015] After the judgment matrix passes the consistency test, the eigenvector of the judgment matrix is ​​calculated and normalized to obtain the parameter level reflection index. No. Parameter weights of the AHP ;

[0016] Step (3.2): Calculate the parameter level reflection index using the entropy weight method The objective weights of each parameter are constructed and the evaluation matrix is ​​normalized to obtain ;

[0017] in, Represents the reflection index after normalization The parameter level evaluation matrix of Indicates the Catenary section reflection index No. The normalized measurement values ​​of the parameters, , m is the number of catenary sections;

[0018] Calculation parameter level reflection indicators No. Information entropy of parameters :

[0019] , ;

[0020] in, Indicates the Catenary section reflection index No. The probability of a parameter appearing;

[0021] Calculate the parameter level reflection index No. Entropy weight method weight of parameters :

[0022] ;

[0023] Step (3.3): Calculate the parameter level reflection index using the combined weight method No. Static composite weights of parameters :

[0024] .

[0025] Furthermore, the method for obtaining the real-time degradation rate of the dynamic and static parameters of the contact network in step (4) is:

[0026] ;

[0027] in, Indicates that the parameter level reflects the index No. Parameters in The real-time degradation rate at the moment, express Time parameter level reflection indicators No. The measured values ​​of the parameters, , n Indicates that the parameter level reflects the index The number of parameters, Indicates the data collection time interval.

[0028] Furthermore, in step (4), based on the real-time degradation rate and the expected degradation rate, an adaptive method is used to dynamically optimize the static composite weights of the parameters at the parameter level in step (3), and the method for obtaining the dynamic composite weight optimization result is:

[0029] ;

[0030] in, Indicates the data collection time, Indicates that the parameter level reflects the index No. Parameters in t The modified weight increment at the moment; Indicated by , , , common N A vector consisting of elements, Indicates the degradation rate difference The first-order backward difference of Indicates that the parameter level reflects the index No. Parameters in The degradation rate difference at each moment, , N is the adaptive method order, N is a positive integer, Indicates that the parameter level reflects the index No. Parameters in The expected degradation rate at time t, Indicates that the parameter level reflects the index No. Parameters in Real-time degradation rate at the moment; Indicated by , , , common N A vector consisting of elements, express The first-order backward difference of Indicates that the parameter level reflects the index No. Parameters in Environmental factors at all times; Indicates that the parameter level reflects the index No. Parameters in The first adaptive gain at time , Indicates that the parameter level reflects the index No. Parameters in The second adaptive gain at time t;

[0031] Parameter level reflection indicators The dynamic composite weight optimization result is: ;

[0032] in, Indicates that the parameter level reflects the index No. Dynamic composite weight optimization results of parameters; Indicates that the parameter level reflects the index No. The static composite weight of the parameters.

[0033] Furthermore, in step (6), a cost function based on the second-order degradation rate differential is constructed and the cost function is solved using the function extreme value method, and the first adaptive gain and the second adaptive gain in step (4) are optimized and updated, including the following steps:

[0034] Step (6.1): Constructing a cost function based on the second-order degradation rate differential :

[0035] ;

[0036] in, is defined as a polynomial, is the pseudo partial derivative variable, express The first-order backward difference of Indicates that the parameter level reflects the index No. Parameters in The real-time degradation rate at the moment, Indicated by , , , common NA vector consisting of elements, express The first-order backward difference of Indicates the expected degradation rate and the real-time degradation rate in The degradation rate difference at each moment, Indicates that the parameter level reflects the index No. Parameters in The expected degradation rate at time t, Indicates that the parameter level reflects the index No. Parameters in Real-time degradation rate at the moment; Indicated by , , , common N A vector consisting of elements, express The first-order backward difference of Indicates that the parameter level reflects the index No. Parameters in Environmental factors at all times; , N is the adaptive method order, N is a positive integer; express The first-order backward difference of is a constant, is a constant, represents the two-norm;

[0037] Step (6.2): ​​Use the function extreme value method to solve the cost function described in step (6.1) and optimize and update the first adaptive gain :

[0038] ;

[0039] in, is the first step length factor, Indicates that the parameter level reflects the index No. Parameters in The first adaptive gain at time t;

[0040] Step (6.3): Constructing a cost function based on the second-order degradation rate differential :

[0041] ;

[0042] in, is defined as a polynomial, express The first-order backward difference of is a constant, is a constant;

[0043] Step (6.4): Use the function extreme value method to solve the cost function described in step (6.3) and optimize and update the second adaptive gain :

[0044] ;

[0045] in, is the second step length factor, Indicates that the parameter level reflects the index No. Parameters in The second adaptive gain at time t.

[0046] Furthermore, the pseudo partial derivative variable The calculation method is:

[0047] ;

[0048] in, for t Pseudo partial derivative variables at time -1, is a constant, is a constant.

[0049] The present invention provides a contact network state assessment system based on adaptive weighted optimization, comprising:

[0050] Catenary status assessment system establishment module: used to establish a catenary status assessment system, the assessment system includes an indicator level and a parameter level, the indicator level includes a safety indicator X1, a smoothness indicator X2, a current-collecting performance indicator X3, and an electrical performance indicator X4, and the parameter level includes several parameters reflecting each indicator level;

[0051] Data processing module: used for collecting dynamic and static parameters of the contact network and processing the dynamic and static parameters of the contact network;

[0052] Initial weight calculation module: used to calculate the subjective and objective weights of each parameter in the parameter level by using the hierarchical analysis method and the entropy weight method respectively, and to calculate the static composite weight of each parameter in the parameter level by using the combined weight method;

[0053] Dynamic weight optimization module: used to collect historical degradation data of the dynamic and static parameters of the contact network, use time series neural network to establish degradation rate prediction models for each parameter of the parameter level, and obtain the expected degradation rate of each parameter through the degradation rate prediction model; calculate the real-time degradation rate of the dynamic and static parameters of the contact network, and based on the real-time degradation rate and the expected degradation rate, use an adaptive method to dynamically optimize the static composite weights of each parameter of the parameter level to obtain a dynamic composite weight optimization result, wherein the adaptive method includes a first adaptive gain and a second adaptive gain;

[0054] Among them, the method for dynamically optimizing the static composite weights of each parameter in the parameter level is:

[0055] ;

[0056] in, Indicates the data collection time, Indicates that the parameter level reflects the index No. Parameters in t The modified weight increment at the moment; , Indicates the degradation rate difference The first-order backward difference of , N is the order of the adaptive method; , express The first-order backward difference of Indicates that the parameter level reflects the index No. Parameters in Environmental factors at all times; Indicates that the parameter level reflects the index No. Parameters in The first adaptive gain at time , Indicates that the parameter level reflects the index No. Parameters in The second adaptive gain at time t;

[0057] Parameter level reflection indicators No. Dynamic composite weight optimization results of parameters for: ;

[0058] in, Indicates that the parameter level reflects the index No. The static composite weight of the parameters;

[0059] Adaptive gain update module: used to construct a cost function based on the second-order degradation rate differential and solve the cost function using the function extreme value method, and optimize and update the first adaptive gain and the second adaptive gain in the dynamic weight optimization module;

[0060] Comprehensive status assessment module: Based on the dynamic composite weight optimization results, the contact network status at the data collection point is assessed.

[0061] Furthermore, the present invention adopts the following technical solutions:

[0062] A non-transitory computer-readable storage medium stores a computer program, which, when executed by a processor, implements the above-mentioned contact network state assessment method based on adaptive weighted optimization.

[0063] Furthermore, the present invention adopts the following technical solutions:

[0064] An electronic device comprises a memory, a processor and a computer program stored in the memory and executable on the processor. When the processor executes the program, the above-mentioned contact network status assessment method based on adaptive weighted optimization is implemented.

[0065] The beneficial technical effects of the present invention are:

[0066] (1) First, the present invention innovatively proposes a weight adaptive optimization mechanism based on dynamic degradation characteristics. By dynamically comparing the real-time degradation rate with the expected degradation rate and combining it with adaptive gain adjustment, the dynamic optimization of weight distribution is achieved, so that the evaluation results can reflect the actual degradation trend of the contact network performance. Secondly, the first adaptive gain and the second adaptive gain are optimized by the second-order degradation rate differential cost function, so that the weight adjustment process has dynamic stability and avoids the oscillation phenomenon of the evaluation results. Finally, the dynamic degradation characteristic parameters of the contact network parameters in the long-term operation process are fully utilized to optimize the weight adjustment, thereby improving the adaptability of the contact network status evaluation.

[0067] (2) Reducing computing resource consumption: The first adaptive gain in the adaptive weight optimization method of the present invention and the second adaptive gain It does not contain any large matrix operations, has low computing resource consumption, and is easy to use in practice;

[0068] (3) Broaden the scope of application: The method and system proposed in the present invention are not only suitable for contact network status assessment, but can also be extended to other status assessment systems, and have broad application prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] Figure 1A contact network status assessment system provided by an embodiment of the present invention;

[0070] Figure 2 A schematic diagram of a process for obtaining dynamic composite weight optimization results in a method for evaluating a contact network state based on adaptive weighted optimization provided by an embodiment of the present invention;

[0071] Figure 3 A schematic diagram of the contact network status assessment system module based on adaptive weighted optimization provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0072] The present invention discloses a contact network state assessment method and system based on adaptive weighted optimization. The method comprises collecting dynamic and static parameters of the contact network, using a hierarchical analysis method and an entropy weight method to calculate the subjective and objective weights of the parameters, and using a combined weight method to calculate the static composite weights. A time series neural network is used to establish a degradation rate prediction model, and current degradation data and environmental change data are obtained in real time. The data are input into the degradation rate prediction model to obtain an expected degradation rate. The real-time degradation rate of the parameters is calculated, and based on the real-time degradation rate and the expected degradation rate, an adaptive method is used to dynamically optimize the static composite weights to obtain dynamic composite weight optimization results. The gain in the adaptive method is optimized and updated. Based on the dynamic composite weight optimization results, the contact network state is assessed until the assessment task is completed. The method and system proposed by the present invention are not only suitable for contact network state assessment, but can also be extended to other state assessment systems, and have broad application prospects.

[0073] The following is a further clear and complete description of the contact network state assessment method and system based on adaptive weighted optimization provided by the present invention in conjunction with the accompanying drawings: Example 1

[0074] Figure 1 The contact network status assessment system provided by this embodiment is given; Figure 2 A schematic diagram of a process for obtaining dynamic composite weight optimization results in a method for evaluating a contact network condition based on adaptive weighted optimization provided in this embodiment is provided. This embodiment provides a method for evaluating a contact network condition based on adaptive weighted optimization, and the method includes the following steps:

[0075] Step (1): Establish a contact network status evaluation system, the evaluation system includes an indicator level and a parameter level, the indicator level includes a safety indicator X1, a smoothness indicator X2, a current receiving performance indicator X3, and an electrical performance indicator X4, the parameter level includes a number of parameters reflecting each indicator level; specifically, the parameter level includes a pull-out value parameter X1 reflecting the safety indicator X1, 11 , guide height parameter X 12 , reflecting the hard point parameter X2 of the smoothness index 21 , Span height parameter X22 , reflecting the contact force parameter X3 of the current-carrying performance index 31 , offline rate parameter X 32 , reflecting the grid voltage parameter X4 of the electrical performance index X 41 , insulation resistance parameter X 42 ;

[0076] Step (2): collecting dynamic and static parameters of the contact network and processing the dynamic and static parameters of the contact network;

[0077] Step (3): Calculate the subjective and objective weights of each parameter at the parameter level using the analytic hierarchy process and entropy weight method respectively, and calculate the static composite weight of each parameter at the parameter level using the combined weight method;

[0078] Step (4): collecting historical degradation data of the dynamic and static parameters of the contact network, using a time series neural network to establish a degradation rate prediction model for each parameter of the parameter level, and obtaining the expected degradation rate of each parameter through the degradation rate prediction model; calculating the real-time degradation rate of the dynamic and static parameters of the contact network, and based on the real-time degradation rate and the expected degradation rate, using an adaptive method to dynamically optimize the static composite weights of each parameter of the parameter level in step (3) to obtain a dynamic composite weight optimization result, wherein the adaptive method includes a first adaptive gain and a second adaptive gain;

[0079] Step (5): Based on the dynamic composite weight optimization results, the state of the contact network at the data collection point is evaluated;

[0080] Step (6): constructing a cost function based on the second-order degradation rate differential and solving the cost function using the function extreme value method, and optimizing and updating the first adaptive gain and the second adaptive gain in step (4);

[0081] Repeat the above steps (4) to (6) to obtain the contact network status evaluation results at the data collection points at different times.

[0082] It should be noted that, according to the contact network state assessment method based on adaptive weighted optimization provided by this embodiment, in step (4), the historical degradation data of the dynamic and static parameters of the contact network are collected, and a degradation rate prediction model of each parameter of the parameter level is established by using a time series neural network. The expected degradation rate of each parameter is obtained through the degradation rate prediction model, which specifically includes: collecting the historical degradation data of the dynamic and static parameters of the contact network, establishing a degradation rate prediction model of each parameter by using a time series neural network, obtaining the current degradation data and environmental factor change data in real time, inputting the degradation rate prediction model, and obtaining the expected degradation rate; the establishment process of the degradation rate prediction model specifically includes: for the indicators reflected in the parameter level For each parameter (pull-out value parameter, guide height parameter, hard point parameter, span height parameter, contact force parameter, offline rate parameter, network voltage parameter, insulation resistance parameter), a dedicated single-variable degradation rate prediction model is established respectively; the historical time series data of the parameter and its corresponding environmental factor data are collected, and after these data are aligned according to a unified time step, a multidimensional training data set containing time series features is constructed; the time series neural network adopts an LSTM-attention mechanism hybrid model, and its input layer receives a multidimensional tensor composed of degradation rate data calculated from the historical time series data of the parameter and corresponding environmental factor data. The bidirectional LSTM layer extracts spatiotemporal features and captures the correlation between parameters, and then uses the attention mechanism to dynamically focus on key time nodes. Finally, the fully connected layer outputs the expected degradation rate of the parameter; in the real-time prediction stage, the real-time degradation rate data and real-time environmental factor data in the current sliding window are input into the trained degradation rate prediction model to output the expected degradation rate of the parameter; of course, when processing the dynamic and static parameters of the contact network in step (2), the parameters are cleaned and normalized to remove noise and outliers.

[0083] In step (3), the subjective and objective weights of each parameter at the parameter level are calculated using the hierarchical analysis method and the entropy weight method respectively, and the static composite weight of each parameter at the parameter level is calculated using the combined weight method, which includes the following steps:

[0084] Step (3.1): Use the hierarchical analysis method to calculate the parameter level reflection index The subjective weight of each parameter is used to quantify the relative importance of the parameters through expert scoring, and a judgment matrix is ​​constructed. ;

[0085] in, Indicates reflection indicators The parameter level judgment matrix, Indicates that the parameter level reflects the index No. Parameters and The relative importance of the parameters, , n Reflect indicators at the parameter level The number of parameters; From the above, we can see that for ,when , Represents the parameter level judgment matrix reflecting the safety index; when , Represents the parameter level judgment matrix reflecting the smoothness index; when , Represents the parameter level judgment matrix reflecting the current receiving performance index; when , Represents a parameter level judgment matrix reflecting electrical performance indicators;

[0086] After the judgment matrix passes the consistency test, the eigenvector of the judgment matrix is ​​calculated and normalized to obtain the parameter level reflection index. No. Parameter weights of the AHP ,in ;

[0087] It should be noted that performing a consistency check on the judgment matrix to determine whether there is a contradiction within the matrix is ​​well known to those skilled in the art. The order of the judgment matrix is ​​less than or equal to 7, and the calculation formula for the consistency check is:

[0088] ;

[0089] ;

[0090] Among them, the order of the judgment matrix here is n That is, the parameter level reflects the index The number of parameters, is the judgment matrix The maximum eigenvalue of CI is the consistency indicator, CR is the random consistency ratio, RI is the average random consistency index, and the average random consistency index of each order judgment matrix is:

[0091]

[0092] when CR <0.1, the judgment matrix passes the consistency test; when CR ≥0.1, the judgment matrix needs to be adjusted in , , optimize the consistency of the judgment matrix until it satisfies CR <0.1;

[0093] For the judgment matrix that passes the consistency test , solve its characteristic equation , and get the maximum eigenvalue The corresponding eigenvector ; For the feature vector To standardize the process, , get the parameter level reflection index No. Parameter weights of the AHP ,in, for No. elements, , ;

[0094] Step (3.2): Calculate the parameter level reflection index using the entropy weight method The objective weights of each parameter are constructed and the evaluation matrix is ​​normalized to obtain ;

[0095] in, Represents the reflection index after normalization The parameter level evaluation matrix of Indicates the Catenary section reflection index No. The normalized measurement values ​​of the parameters, , , m is the number of catenary sections; from the above, we can see that for ,when , hour, represents the parameter level evaluation matrix reflecting the safety index after normalization; when , hour, represents the parameter level evaluation matrix reflecting the smoothness index after normalization; when , hour, represents the parameter level evaluation matrix reflecting the current collecting performance index after normalization; when , hour, It represents the parameter level evaluation matrix reflecting the electrical performance index after normalization;

[0096] Calculation parameter level reflection indicators No. Information entropy of parameters :

[0097] , ;

[0098] in, Indicates the Catenary section reflection index No. The probability of a parameter appearing;

[0099] Calculate the parameter level reflection index No. Entropy weight method weight of parameters :

[0100] ;

[0101] Step (3.3): Calculate the parameter level reflection index using the combined weight method No. Static composite weights of parameters :

[0102] .

[0103] In addition, the method for obtaining the real-time degradation rate of the dynamic and static parameters of the contact network in step (4) is:

[0104] ;

[0105] in, Indicates that the parameter level reflects the index No. Parameters in The real-time degradation rate at the moment, express Time parameter level reflection indicators No. The measured values ​​of the parameters should be noted that these parameter values ​​are obtained by dynamic or static measurement of the contact network detection equipment, specifically: the pull-out value parameter X 11 , guide height parameter X 12 It is directly measured by a contact network geometric parameter detection device (such as a laser measuring instrument or an image recognition system); the hard point parameter X 21 , Span height parameter X 22 It is calculated based on the pantograph acceleration sensor or the contact line height continuous scanning data; the contact force parameter X 31 , offline rate parameter X 32 The grid voltage parameter X is collected in real time by the force sensor and arc detection device carried by the pantograph. 41 , insulation resistance parameter X 42 It is directly measured using on-board voltage transformers and insulation monitoring equipment; , n Indicates that the parameter level reflects the index The number of parameters, Indicates the data collection time interval.

[0106] Specifically, in step (4), based on the real-time degradation rate and the expected degradation rate, an adaptive method is used to dynamically optimize the static composite weights of the parameters at the parameter level in step (3), and the method for obtaining the dynamic composite weight optimization result is:

[0107] ;

[0108] in, Indicates the data collection time, Indicates that the parameter level reflects the index No. Parameters in The modified weight increment at the moment; Indicated by , , , common N A vector consisting of elements, Indicates the degradation rate difference The first-order backward difference of Indicates that the parameter level reflects the index No. Parameters in The degradation rate difference at each moment, , N is the adaptive method order, N is a positive integer, Indicates that the parameter level reflects the index No. Parameters in The expected degradation rate at time t, Indicates that the parameter level reflects the index No. Parameters in Real-time degradation rate at the moment; Indicated by , , , common N A vector consisting of elements, express The first-order backward difference of Indicates that the parameter level reflects the index No. Parameters in Environmental factors at all times; Indicates that the parameter level reflects the index No. Parameters in The first adaptive gain at time , Indicates that the parameter level reflects the index No. Parameters in The second adaptive gain at time t;

[0109] Parameter level reflection indicators No. The dynamic composite weight optimization results of the parameters are: ;

[0110] in, Indicates that the parameter level reflects the index No. Dynamic composite weight optimization results of parameters; Indicates that the parameter level reflects the index No. The static composite weight of the parameters.

[0111] When the static composite weight is dynamically optimized by the adaptive method in step (4), the first adaptive gain and the second adaptive gain It is a key parameter. If the gain is too large, the weight may be overly sensitive to short-term fluctuations in the degradation rate. If the gain is too small, the real degradation trend cannot be reflected in time. In order to balance the influence of historical and real-time data, the gain value needs to be dynamically adjusted to adapt to the degradation characteristics under different circumstances. Specifically, in step (6), a cost function based on the second-order degradation rate differential is constructed and the function extreme value method is used to solve the cost function, and the first adaptive gain and the second adaptive gain in step (4) are optimized and updated, including the following steps:

[0112] Step (6.1): Constructing a cost function based on the second-order degradation rate differential :

[0113] ;

[0114] in, is defined as a polynomial, is the pseudo partial derivative variable, express The first-order backward difference of Indicates that the parameter level reflects the index No. Parameters in The real-time degradation rate at the moment, Indicated by , , , common N A vector consisting of elements, express The first-order backward difference of Indicates the expected degradation rate and the real-time degradation rate in The degradation rate difference at each moment, Indicates that the parameter level reflects the index No. Parameters in The expected degradation rate at time t, Indicates that the parameter level reflects the index No. Parameters in Real-time degradation rate at the moment; Indicated by , , , common N A vector consisting of elements, express The first-order backward difference of Indicates that the parameter level reflects the index No. Parameters in Environmental factors at all times; , N is the adaptive method order, N is a positive integer; express The first-order backward difference of is a constant, is a constant, represents the two-norm;

[0115] Step (6.2): ​​Use the function extreme value method to solve the cost function described in step (6.1) and optimize and update the first adaptive gain :

[0116] ;

[0117] in, is the first step length factor, Indicates that the parameter level reflects the index No. Parameters in The first adaptive gain at time t;

[0118] Step (6.3): Constructing a cost function based on the second-order degradation rate differential :

[0119] ;

[0120] in, is defined as a polynomial, express The first-order backward difference of is a constant, is a constant;

[0121] Step (6.4): Use the function extreme value method to solve the cost function described in step (6.3) and optimize and update the second adaptive gain :

[0122] ;

[0123] in, is the second step length factor, Indicates that the parameter level reflects the index No. Parameters in The second adaptive gain at time t.

[0124] Among them, the pseudo partial derivative variable The calculation method is:

[0125] ;

[0126] in, for t Pseudo partial derivative variables at time -1, is a constant, is a constant. Example 2

[0127] This embodiment provides a contact network status assessment system based on adaptive weighted optimization. Figure 3 A schematic diagram of the module connection of the contact network status assessment system based on adaptive weighted optimization provided in this embodiment is provided, including:

[0128] Catenary status assessment system establishment module: used to establish a catenary status assessment system, the assessment system includes an indicator level and a parameter level, the indicator level includes a safety indicator X1, a smoothness indicator X2, a current-collecting performance indicator X3, and an electrical performance indicator X4, and the parameter level includes several parameters reflecting each indicator level;

[0129] Data processing module: used for collecting dynamic and static parameters of the contact network and processing the dynamic and static parameters of the contact network;

[0130] Initial weight calculation module: used to calculate the subjective and objective weights of each parameter in the parameter level by using the hierarchical analysis method and the entropy weight method respectively, and to calculate the static composite weight of each parameter in the parameter level by using the combined weight method;

[0131] Dynamic weight optimization module: used to collect historical degradation data of the dynamic and static parameters of the contact network, use time series neural network to establish degradation rate prediction models for each parameter of the parameter level, and obtain the expected degradation rate of each parameter through the degradation rate prediction model; calculate the real-time degradation rate of the dynamic and static parameters of the contact network, and based on the real-time degradation rate and the expected degradation rate, use an adaptive method to dynamically optimize the static composite weights of each parameter of the parameter level to obtain a dynamic composite weight optimization result, wherein the adaptive method includes a first adaptive gain and a second adaptive gain;

[0132] Among them, the method for dynamically optimizing the static composite weights of each parameter in the parameter level is:

[0133] ;

[0134] in, Indicates the data collection time, Indicates that the parameter level reflects the index No. Parameters in The modified weight increment at the moment; , Indicates the degradation rate difference The first-order backward difference of , N is the order of the adaptive method; , express The first-order backward difference of Indicates that the parameter level reflects the index No. Parameters in Environmental factors at all times; Indicates that the parameter level reflects the index No. Parameters in The first adaptive gain at time , Indicates that the parameter level reflects the index No. Parameters in The second adaptive gain at time t;

[0135] Parameter level reflection indicators No. Dynamic composite weight optimization results of parameters for: ;

[0136] in, Indicates that the parameter level reflects the index No. The static composite weight of the parameters;

[0137] Adaptive gain update module: used to construct a cost function based on the second-order degradation rate differential and solve the cost function using the function extreme value method, and optimize and update the first adaptive gain and the second adaptive gain in the dynamic weight optimization module;

[0138] Comprehensive status assessment module: Based on the dynamic composite weight optimization results, the contact network status at the data collection point is assessed.

[0139] Furthermore, this embodiment provides the following technical solutions:

[0140] A non-transitory computer-readable storage medium stores a computer program, which, when executed by a processor, implements the above-mentioned contact network state assessment method based on adaptive weighted optimization.

[0141] Furthermore, the present invention adopts the following technical solutions:

[0142] An electronic device comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method for evaluating the contact network status based on adaptive weighted optimization as described above is implemented.

[0143] From the above description of the embodiments, it will be clear to those skilled in the art that the facilities of the present invention can be implemented using software plus the necessary general-purpose hardware platforms. Embodiments of the present invention can be implemented using existing processors, or by dedicated processors used in appropriate systems for this or other purposes, or by hardwired systems. Embodiments of the present invention also include non-transitory computer-readable storage media, including machine-readable media for carrying or having stored thereon machine-executable instructions or data structures. Such machine-readable media can be any available medium that can be accessed by a general-purpose or special-purpose computer or other machine with a processor. For example, such machine-readable media can include RAM, ROM, EPROM, EEPROM, CD-ROM or other optical disk storage, magnetic disk storage or other magnetic storage devices, or any other medium that can be used to carry or store the required program code in the form of machine-executable instructions or data structures and that can be accessed by a general-purpose or special-purpose computer or other machine with a processor. When information is transmitted or provided to a machine via a network or other communications connection (hardwired, wireless, or a combination of hardwired and wireless), such connection is also considered a machine-readable medium.

[0144] Thus far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art may make equivalent changes or substitutions to the relevant technical features, and the technical solutions after such changes or substitutions will fall within the scope of protection of the present invention.

Claims

1. A method for evaluating the state of a contact network based on adaptive weighted optimization, characterized in that: The method comprises the following steps: Step (1): establishing a contact network status evaluation system, wherein the evaluation system includes an indicator level and a parameter level, wherein the indicator level includes a safety indicator X1, a smoothness indicator X2, a current collecting performance indicator X3, and an electrical performance indicator X4, and the parameter level includes a number of parameters reflecting each indicator level; Step (2): collecting dynamic and static parameters of the contact network and processing the dynamic and static parameters of the contact network; Step (3): Calculate the subjective and objective weights of each parameter at the parameter level using the analytic hierarchy process and entropy weight method respectively, and calculate the static composite weight of each parameter at the parameter level using the combined weight method; Step (4): collecting historical degradation data of the dynamic and static parameters of the contact network, using a time series neural network to establish a degradation rate prediction model for each parameter of the parameter level, and obtaining the expected degradation rate of each parameter through the degradation rate prediction model; calculating the real-time degradation rate of the dynamic and static parameters of the contact network, and based on the real-time degradation rate and the expected degradation rate, using an adaptive method to dynamically optimize the static composite weights of each parameter of the parameter level in step (3) to obtain a dynamic composite weight optimization result, wherein the adaptive method includes a first adaptive gain and a second adaptive gain; Among them, the method for dynamically optimizing the static composite weights of each parameter in the parameter level is: ； in, Indicates the data collection time, Indicates that the parameter level reflects the index No. Parameters in t The modified weight increment at the moment; , Indicates the degradation rate difference The first-order backward difference of , N is the order of the adaptive method; , express The first-order backward difference of Indicates that the parameter level reflects the index No. Parameters in Environmental factors at all times; Indicates that the parameter level reflects the index No. Parameters in The first adaptive gain at time , Indicates that the parameter level reflects the index No. Parameters in The second adaptive gain at time t; Parameter level reflection indicators No. Dynamic composite weight optimization results of parameters for: ; in, Indicates that the parameter level reflects the index No. The static composite weight of the parameters; Step (5): Based on the dynamic composite weight optimization results, the state of the contact network at the data collection point is evaluated; Step (6): constructing a cost function based on the second-order degradation rate differential and solving the cost function using the function extreme value method, and optimizing and updating the first adaptive gain and the second adaptive gain in step (4); Repeat the above steps (4) to (6) to obtain the contact network status evaluation results at the data collection points at different times.

2. The method for evaluating contact network status based on adaptive weighted optimization according to claim 1, characterized in that: In step (3), the subjective and objective weights of each parameter at the parameter level are calculated using the hierarchical analysis method and the entropy weight method respectively, and the static composite weight of each parameter at the parameter level is calculated using the combined weight method, which includes the following steps: Step (3.1): Use the hierarchical analysis method to calculate the parameter level reflection index The subjective weight of each parameter is used to quantify the relative importance of the parameters through expert scoring, and a judgment matrix is ​​constructed. ; in, Indicates reflection indicators The parameter level judgment matrix, Indicates that the parameter level reflects the index No. Parameters and The relative importance of the parameters, , n Reflect indicators at the parameter level The number of parameters; After the judgment matrix passes the consistency test, the eigenvector of the judgment matrix is ​​calculated and normalized to obtain the parameter level reflection index. No. Parameter weights of the AHP ; Step (3.2): Calculate the parameter level reflection index using the entropy weight method The objective weights of each parameter are constructed and the evaluation matrix is ​​normalized to obtain ; in, Represents the reflection index after normalization The parameter level evaluation matrix of Indicates the Catenary section reflection index No. The normalized measurement values ​​of the parameters, , m is the number of catenary sections; Calculation parameter level reflection indicators No. Information entropy of parameters : ， ； in, Indicates the Catenary section reflection index No. The probability of a parameter appearing; Calculate the parameter level reflection index No. Entropy weight method weight of parameters : ； Step (3.3): Calculate the parameter level reflection index using the combined weight method No. Static composite weights of parameters : 。 3. The method for evaluating contact network status based on adaptive weighted optimization according to claim 1, characterized in that: The method for obtaining the real-time degradation rate of the dynamic and static parameters of the contact network in step (4) is: ； in, Indicates that the parameter level reflects the index No. Parameters in The real-time degradation rate at the moment, express Time parameter level reflection indicators No. The measured values ​​of the parameters, , n Indicates that the parameter level reflects the index The number of parameters, Indicates the data collection time interval.

4. The method for evaluating contact network status based on adaptive weighted optimization according to claim 1, characterized in that: In step (6), a cost function based on the second-order degradation rate differential is constructed and the cost function is solved using the function extreme value method, and the first adaptive gain and the second adaptive gain in step (4) are optimized and updated, including the following steps: Step (6.1): Constructing a cost function based on the second-order degradation rate differential : ； in, is defined as a polynomial, is the pseudo partial derivative variable, express The first-order backward difference of Indicates that the parameter level reflects the index No. Parameters in The real-time degradation rate at the moment, Indicated by , , , common N A vector consisting of elements, express The first-order backward difference of Indicates the expected degradation rate and the real-time degradation rate in The degradation rate difference at each moment, Indicates that the parameter level reflects the index No. Parameters in The expected degradation rate at time t, Indicates that the parameter level reflects the index No. Parameters in Real-time degradation rate at the moment; Indicated by , , , common N A vector consisting of elements, express The first-order backward difference of Indicates that the parameter level reflects the index No. Parameters in Environmental factors at all times; , N is the adaptive method order, N is a positive integer; express The first-order backward difference of is a constant, is a constant, represents the two-norm; Step (6.2): ​​Use the function extreme value method to solve the cost function described in step (6.1) and optimize and update the first adaptive gain : ； in, is the first step length factor, Indicates that the parameter level reflects the index No. Parameters in The first adaptive gain at time t; Step (6.3): Constructing a cost function based on the second-order degradation rate differential : ； in, is defined as a polynomial, express The first-order backward difference of is a constant, is a constant; Step (6.4): Use the function extreme value method to solve the cost function described in step (6.3) and optimize and update the second adaptive gain : ； in, is the second step length factor, Indicates that the parameter level reflects the index No. Parameters in The second adaptive gain at time t.

5. The method for evaluating contact network status based on adaptive weighted optimization according to claim 4, characterized in that: The pseudo partial derivative variable The calculation method is: ； in, for t Pseudo partial derivative variables at time -1, is a constant, is a constant.

6. The method for evaluating contact network status based on adaptive weighted optimization according to claim 1, characterized in that: In step (2), the dynamic and static parameters of the contact network are processed, including cleaning and normalizing the parameters to remove noise and abnormal values.

7. The contact network status assessment system based on adaptive weighted optimization is characterized by: include: Catenary status assessment system establishment module: used to establish a catenary status assessment system, the assessment system includes an indicator level and a parameter level, the indicator level includes a safety indicator X1, a smoothness indicator X2, a current-collecting performance indicator X3, and an electrical performance indicator X4, and the parameter level includes several parameters reflecting each indicator level; Data processing module: used for collecting dynamic and static parameters of the contact network and processing the dynamic and static parameters of the contact network; Initial weight calculation module: used to calculate the subjective and objective weights of each parameter in the parameter level by using the hierarchical analysis method and the entropy weight method respectively, and to calculate the static composite weight of each parameter in the parameter level by using the combined weight method; Dynamic weight optimization module: used to collect historical degradation data of the dynamic and static parameters of the contact network, use time series neural network to establish degradation rate prediction models for each parameter of the parameter level, and obtain the expected degradation rate of each parameter through the degradation rate prediction model; calculate the real-time degradation rate of the dynamic and static parameters of the contact network, and based on the real-time degradation rate and the expected degradation rate, use an adaptive method to dynamically optimize the static composite weights of each parameter of the parameter level to obtain a dynamic composite weight optimization result, wherein the adaptive method includes a first adaptive gain and a second adaptive gain; Among them, the method for dynamically optimizing the static composite weights of each parameter in the parameter level is: ； in, Indicates the data collection time, Indicates that the parameter level reflects the index No. Parameters in t The modified weight increment at the moment; , Degradation rate difference The first-order backward difference of , N is the order of the adaptive method; , express The first-order backward difference of Indicates that the parameter level reflects the index No. Parameters in Environmental factors at all times; Indicates that the parameter level reflects the index No. Parameters in The first adaptive gain at time , Indicates that the parameter level reflects the index No. Parameters in The second adaptive gain at time t; Parameter level reflection indicators No. Dynamic composite weight optimization results of parameters for: ; in, Indicates that the parameter level reflects the index No. The static composite weight of the parameters; Adaptive gain update module: used to construct a cost function based on the second-order degradation rate differential and solve the cost function using the function extreme value method, and optimize and update the first adaptive gain and the second adaptive gain in the dynamic weight optimization module; Comprehensive status assessment module: Based on the dynamic composite weight optimization results, the contact network status at the data collection point is assessed.

8. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for evaluating the state of an overhead contact network based on adaptive weighted optimization as described in any one of claims 1 to 6 is implemented.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the contact network status assessment method based on adaptive weighted optimization as described in any one of claims 1 to 6 is implemented.

Images