Railway operations communication network optimization system fused with power grid information

Abstract

The application relates to the technical field of railway operation and maintenance communication and discloses a railway operation and maintenance communication network optimization system fusing power grid information. The system comprises a multi-source data fusion module, which collects power, railway and environmental data to generate a time-space multi-dimensional data cube; a communication quality evaluation module, which analyzes data by using a fuzzy logic fusion algorithm to obtain a communication link quality degradation index sequence; a network topology optimization module, which constructs a dynamic Bayesian network model to calculate an optimal communication routing table; a dynamic resource scheduling module, which designs a double-layer planning algorithm to realize joint scheduling of frequency spectrum and power; and an abnormality cooperative processing module, which establishes a federal learning framework to generate a cross-domain cooperative repair instruction set. The system can accurately evaluate communication quality, optimize network topology, reasonably schedule resources and cooperatively process abnormalities by fusing power grid information, so that the performance and reliability of the railway operation and maintenance communication network are effectively improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of railway operation and maintenance communication technology, in particular to a railway operation and maintenance communication network optimization system integrated with power grid information. BACKGROUND

[0002] In the railway transportation system, the stable operation of the railway operation and maintenance communication network is of great importance. It not only undertakes the transmission tasks of key information such as train operation scheduling and equipment state monitoring, but also is closely related to the safe operation of the railway. However, the current railway operation and maintenance communication network faces many challenges in actual operation.

[0003] On the one hand, there is a complex interaction between the communication network and the power grid environment. The fluctuation of power load can produce electromagnetic interference, affecting the quality of communication signals. For example, during the peak period of electricity consumption, the harmonic content of the power system increases, and these harmonics can be coupled into the communication line, causing communication signal errors, attenuation and other problems, and even causing communication interruption in severe cases. At the same time, changes in the topology of the power grid can also cause instability in the communication network. When the power grid is undergoing equipment maintenance, fault isolation and other operations, it will change the electromagnetic environment of the power grid, thereby interfering with the transmission path and quality of the communication signal.

[0004] On the other hand, the traditional resource scheduling mode of the railway operation and maintenance communication network is relatively extensive. In terms of spectrum resource allocation, it is often based on fixed rules or experience, without fully considering the dynamic changes of real-time load and communication demand of the power grid. This leads to the fact that in some periods or regions, spectrum resources are either idle and wasted or insufficiently allocated, failing to meet the actual demand of communication services. In terms of power allocation, there is also a lack of effective balance management of base station energy consumption, and some base stations may be damaged prematurely due to long-term high-power operation, increasing maintenance costs and operational risks.

[0005] In addition, when the railway operation and maintenance communication network encounters an abnormality, the existing handling mechanism has obvious deficiencies. Communication failures and power grid failures are often interrelated, but there is currently a lack of effective means to simultaneously perceive and cooperatively handle these two types of failures. Each system is independent of the others and cannot quickly share fault information, resulting in low efficiency in handling faults and long recovery time, which seriously affects the normal order of railway transportation. Moreover, with the continuous expansion of railway transportation business, such as the popularization of high-speed trains and the development of intelligent transportation systems, the performance requirements of the railway operation and maintenance communication network are becoming higher and higher. The existing communication network optimization technology is difficult to adapt to these new demands and cannot achieve deep integration of communication network and power grid information, making it difficult to ensure communication quality, improve resource utilization and cope with complex faults. SUMMARY

[0006] The present application aims to provide a railway operation communication network optimization system integrated with power grid information to solve the problems raised in the background.

[0007] To achieve the above-mentioned purpose, the present application provides the following technical solutions: a railway operation communication network optimization system integrated with power grid information, the system comprising:

[0008] A multi-source data fusion module is configured to synchronously collect power load fluctuations, power grid topology states, rail vibration frequency spectra, communication signal strengths and environmental meteorological data through heterogeneous sensors, and generate a spatio-temporal multi-dimensional data cube;

[0009] A communication quality evaluation module is configured to perform nonlinear correlation analysis on the spatio-temporal multi-dimensional data cube using a fuzzy logic fusion algorithm, and generate a communication link quality degradation index sequence;

[0010] A network topology optimization module is configured to construct a path planning model based on a dynamic Bayesian network, and calculate an optimal communication routing table under power interference conditions according to the degradation index sequence;

[0011] A dynamic resource scheduling module is configured to design a resource allocation algorithm based on double-layer planning, and generate a joint scheduling strategy of spectrum and power in combination with real-time power grid load and communication demand prediction values;

[0012] An abnormality cooperative processing module is configured to establish a distributed federated learning framework, and generate a cross-domain cooperative repair instruction set according to power grid fault features and communication interruption modes.

[0013] Preferably, the implementation of the multi-source data fusion module comprises:

[0014] The heterogeneous data streams are subjected to spatio-temporal stamp alignment processing, and the sampling rate differences of the sensors are eliminated through a sliding window mechanism;

[0015] A fuzzy logic rule base is constructed, and the power grid harmonic distortion rate and the communication bit error rate are mapped into membership vectors;

[0016] An evidence theory synthesis algorithm is used to weight the reliability of conflicting data, and a reduced fusion feature matrix is output.

[0017] Preferably, the construction of the communication quality evaluation module comprises:

[0018] The input variables of the fuzzy inference system are defined as signal attenuation gradient, multipath time delay spread and power grid harmonic injection energy;

[0019] Gaussian membership functions are defined to depict the degradation trends of the variables, and a three-dimensional rule surface is generated;

[0020] Key degradation modes are extracted through a lambda-cut set method, and a communication link quality level classification result is output.

[0021] Preferably, the design of the network topology optimization module comprises:

[0022] Mapping the substation nodes and communication relay stations as a two-layer node structure of the Bayesian network;

[0023] Calculating the conditional probability distribution table of the communication bit error rate under the power grid flow mutation event, using the Markov chain Monte Carlo sampling algorithm to solve the maximum posterior probability path, and generating an anti-interference routing strategy.

[0024] Preferably, the implementation of the dynamic resource scheduling module comprises:

[0025] Establishing an upper model to minimize the power grid harmonic interference as the target, optimizing the communication frequency band selection strategy, and constructing a lower model to balance the base station energy consumption as the constraint, and solving the power allocation Pareto frontier;

[0026] Using the alternating direction multiplier method to iteratively solve the double-layer model, and outputting the spectrum-power joint configuration scheme.

[0027] Preferably, the establishment of the abnormality cooperative processing module comprises:

[0028] Training a lightweight fault diagnosis model at the local node, extracting the hidden variable relationship between the power grid transient characteristics and the communication packet loss mode; designing a model aggregation protocol to fuse the gradient update amount of each node through a differential privacy protection mechanism;

[0029] Generating a cooperative repair decision tree containing fault isolation instructions and channel switching strategies.

[0030] Preferably, the network topology optimization module further comprises:

[0031] Building a real-time topology update mechanism to capture the transient change characteristics of the power grid topology through a sliding time window;

[0032] Defining a prediction correction algorithm based on Kalman filtering to dynamically correct the path delay estimation value in the communication routing table, using a tabu search strategy to quickly locate a suboptimal routing scheme in a preset solution space, and generating a dynamic path switching instruction against sudden interference.

[0033] Preferably, the system further comprises:

[0034] Building a spatiotemporal correlation database and using a tensor decomposition algorithm to mine cross-domain association rules of historical power grid events and communication failures;

[0035] Defining a rule engine to match the current data pattern in real time and generating a preventive maintenance suggestion queue.

[0036] Preferably, the system further comprises:

[0037] Deploy edge computing node cluster, use stream computing framework to realize end-to-end delay compression of data collection and decision generation;

[0038] Define priority scheduling algorithm, dynamically allocate computing resources to process optimization tasks with different urgency.

[0039] Preferably, the application also includes a railway operation communication network optimization method integrating power grid information, comprising the following steps:

[0040] Step 1: Synchronously collect power load fluctuation, power grid topology state, rail vibration spectrum, communication signal strength and environmental meteorological data through heterogeneous sensors to generate a spatio-temporal multi-dimensional data cube;

[0041] Step 2: Perform nonlinear correlation analysis on the spatio-temporal multi-dimensional data cube using fuzzy logic fusion algorithm to generate a communication link quality degradation index sequence;

[0042] Step 3: Construct a path planning model based on dynamic Bayesian network, and calculate the optimal communication routing table under power disturbance conditions according to the degradation index sequence;

[0043] Step 4: Design a resource allocation algorithm based on double-layer planning, and generate a joint scheduling strategy of spectrum and power combining real-time load of power grid and communication demand prediction value;

[0044] Step 5: Establish a distributed federated learning framework to generate cross-domain collaborative repair instruction set according to power grid fault characteristics and communication interruption mode.

[0045] Compared with the prior art, the application has the following advantages:

[0046] From the perspective of data fusion, the multi-source data fusion module synchronously collects multiple types of data such as power load fluctuation and power grid topology state through heterogeneous sensors, and generates a spatio-temporal multi-dimensional data cube. This process realizes the organic integration of data from different fields, providing a comprehensive data foundation for subsequent accurate analysis and decision-making. It overcomes the limitations of traditional systems that rely only on a single data type for analysis, enabling the system to more comprehensively and accurately understand the operation status of the railway operation communication network and the power grid. For example, by fusing power grid harmonic distortion rate and communication bit error rate data, it can deeply explore the potential relationship between the two, providing strong support for communication quality evaluation and fault prediction.

[0047] In terms of communication quality assessment, a fuzzy logic fusion algorithm is used to generate a sequence of communication link quality degradation indices. This method fully considers the nonlinear relationships between various influencing factors such as signal attenuation gradient, multipath time delay spread, and power grid harmonic injection energy. Compared to traditional simple evaluation methods, it can more accurately reflect the actual quality of the communication link. This helps to timely discover potential problems in the communication link, take measures to optimize and maintain in advance, reduce the probability of communication failure, and ensure the stability and reliability of railway communication.

[0048] The network topology optimization module constructs a path planning model based on a dynamic Bayesian network to calculate the optimal communication routing table under power interference conditions. This enables the communication network to dynamically adjust the routing according to the real-time state of the power grid, effectively avoiding areas of communication quality decline caused by power interference, and improving the anti-interference capability of communication. Even in the event of sudden conditions such as power flow changes, the system can quickly find the optimal or sub-optimal routing to ensure the continuity of communication and avoid the impact of communication interruption on railway transportation safety.

[0049] The dynamic resource scheduling module designs a resource allocation algorithm based on double-layer planning, combining real-time load of the power grid and communication demand prediction values to generate a joint scheduling strategy for frequency spectrum and power. On the one hand, the upper model optimizes communication frequency band selection to minimize power harmonic interference; on the other hand, the lower model solves power allocation with the constraint of balancing base station energy consumption, improving resource utilization efficiency and reducing operating costs. For example, under the premise of meeting communication demand, reasonable allocation of frequency spectrum and power resources avoids overuse of some frequency bands and base stations, extending the service life of equipment.

[0050] The abnormality cooperative processing module establishes a distributed federated learning framework to generate a cross-domain cooperative repair instruction set. This module breaks down the information barriers between the communication system and the power system, enabling cooperative analysis and processing of power grid fault characteristics and communication interruption patterns. When an anomaly occurs, it can quickly develop and execute cross-domain cooperative repair strategies, such as isolating fault areas and switching communication channels, significantly reducing fault processing time and improving the reliability and stability of the railway operation communication network.

[0051] In addition, the spatiotemporal correlation database and rule engine built by the system can mine cross-domain association rules between historical power grid events and communication failures, generating preventive maintenance recommendations. This helps to identify potential fault risks in advance, take preventive measures, and reduce the frequency and impact of faults. The deployment of edge computing node clusters and priority scheduling algorithms achieves end-to-end latency compression for data collection and decision generation, enabling rapid response to emergency optimization tasks and further improving the real-time performance and overall performance of the system. BRIEF DESCRIPTION OF DRAWINGS

[0052] Figure 1The working principle diagram of the railway operation and maintenance communication network optimization system fusing power grid information according to the present application;

[0053] Figure 2 The working principle diagram of the communication quality evaluation module;

[0054] Figure 3 The working principle diagram of the dynamic resource scheduling module;

[0055] Figure 4 The working principle diagram of the network topology optimization module extension function. DETAILED DESCRIPTION

[0056] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative work fall within the protection scope of the present application.

[0057] Please refer to Figures 1-4 The present application provides a railway operation and maintenance communication network optimization system fusing power grid information, and the overall implementation scheme is as follows:

[0058] Multi-source data fusion: synchronous acquisition of power load fluctuation, power grid topology state, rail vibration spectrum, communication signal strength and environmental meteorological data by using heterogeneous sensors, and then generation of a spatio-temporal multi-dimensional data cube. These heterogeneous sensors are distributed along the railway and around the related power facilities, and can obtain various key data in real time.

[0059] Communication quality evaluation: nonlinear correlation analysis of the generated spatio-temporal multi-dimensional data cube by using a fuzzy logic fusion algorithm, and finally generation of a communication link quality degradation index sequence, so as to accurately evaluate the quality status of the communication link.

[0060] Network topology optimization: construction of a path planning model based on a dynamic Bayesian network, calculation of an optimal communication routing table under the condition of power interference according to the communication link quality degradation index sequence, so as to realize optimization of the network topology and guarantee the stability of the communication.

[0061] Dynamic resource scheduling: design of a resource allocation algorithm based on double-layer planning, combination of the real-time load of the power grid and the predicted value of the communication demand, generation of a joint scheduling strategy of spectrum and power, and realization of reasonable allocation of communication resources.

[0062] Abnormal collaborative processing: establishment of a distributed federal learning framework, generation of a cross-domain collaborative repair instruction set according to the power grid fault features and the communication interruption mode, so as to quickly and effectively process the abnormal situation when it occurs.

[0063] The implementation of the present application is further illustrated below in connection with Examples 1 to 5.

[0064] Example 1:

[0065] This example is directed to the specific implementation of the multi-source data fusion module. In actual application scenarios, a large number of sensors of different types are distributed along the railway, such as sensors for monitoring power load, devices for monitoring power grid topology state, sensors for acquiring rail vibration spectrum, devices for detecting communication signal strength, and instruments for collecting environmental meteorological data, etc. Due to differences in manufacturers, models, and functions, the output data streams of these sensors have problems of inconsistent time-space stamps and different sampling rates.

[0066] To solve these problems, first, the heterogeneous data streams are subjected to time-space stamp alignment processing. By adding accurate time labels to each sensor data and arranging them in chronological order, the consistency of data from different sources in the time dimension is ensured. At the same time, the sliding window mechanism is used to eliminate the differences in sensor sampling rates. The size of the sliding window is set according to the actual data situation and system performance requirements, for example, set to N sampling points. During the sliding process, interpolation or decimation operations are performed on data of different sampling rates to make them fused on the same time scale.

[0067] Constructing a fuzzy logic rule base is one of the key steps of multi-source data fusion. In actual operation, there is a complex nonlinear relationship between power grid harmonic distortion rate and communication error rate. Let the power grid harmonic distortion rate be denoted as and the communication error rate be denoted as Through a large number of experimental data and expert experience, a fuzzy logic rule is established to map and to membership vectors. For example, define fuzzy subsets as "low", "medium", and "high", and determine the membership degrees of and to each fuzzy subset according to their values.

[0068] The evidence theory synthesis algorithm is used to weight the credibility of conflicting data. Suppose there are two evidence sources and , and their credibility for a certain proposition is and respectively (where is a certain proposition in the proposition set), then according to the evidence theory synthesis formula

[0069]

[0070] the fused credibility In the system, the measurements of the same physical quantity by different sensors may conflict. The algorithm can effectively improve the reliability of the data by processing the conflicting data. Finally, the reduced fusion feature matrix is output, providing high-quality data support for subsequent communication quality evaluation and other operations.

[0071] Embodiment 2

[0072] This embodiment focuses on the construction process of the communication quality evaluation module. In the railway operation communication network, the quality of the communication link is affected by many factors, such as signal attenuation gradient, multipath delay spread, and power grid harmonic injection energy. Therefore, the input variables of the fuzzy inference system are defined as signal attenuation gradient , multipath delay spread , and power grid harmonic injection energy .

[0073] In order to accurately depict the degradation trend of each variable, Gaussian membership function is used. The expression of Gaussian membership function is , where is the variable value, is the center value of the membership function, is the standard deviation. For example, for signal attenuation gradient , the center value and the standard deviation of the membership function are determined according to historical data and actual experience. When the signal attenuation gradient is , its membership degree is .

[0074] Similarly, the corresponding Gaussian membership functions of multipath delay spread and power grid harmonic injection energy are also determined.

[0075] Through the membership functions of the three variables, a three-dimensional rule surface is generated. In actual operation, according to a large amount of experimental data and communication theory, fuzzy rules are formulated, such as when the signal attenuation gradient is "high", the multipath delay spread is "large", and the power grid harmonic injection energy is "strong", the communication link quality is "poor". Apply these rules to the three-dimensional space to form a three-dimensional rule surface.

[0076] Finally, the key degradation mode is extracted by λ-cut method. λ-cut method is a commonly used method in fuzzy sets. Let the fuzzy set be , its λ-cut , where is the of membership. In this embodiment, by adjusting the value of λ, the key degradation mode is extracted from the three-dimensional regular curved surface, and the communication link quality level classification result is output, which divides the communication link quality into different levels such as "excellent", "good", "medium" and "poor", providing an important basis for subsequent network optimization.

[0077] Embodiment 3:

[0078] This embodiment focuses on the design details of the network topology optimization module. In the scenario of the integration of railway operation communication network and power grid, the substation node and the communication relay station are mapped into the double-layer node structure of the Bayesian network. Among them, the substation node mainly reflects the operation state of the power grid, and the communication relay station node is responsible for the transmission of communication signals. In the Bayesian network, there is a probability dependency relationship between nodes.

[0079] Calculating the conditional probability distribution table of the communication bit error rate under the power grid flow mutation event is an important part of network topology optimization. Assuming that the power grid flow mutation event is , the communication bit error rate is , and the conditional probability distribution is obtained through analysis and statistics of a large amount of historical data. For example, when the power grid flow mutation degree is , the probability of the communication bit error rate being is .

[0080] The Markov Chain Monte Carlo sampling algorithm is used to solve the maximum a posteriori probability path. The Markov Chain Monte Carlo sampling algorithm is a method of sampling in the probability space. By constructing a Markov chain, its stationary distribution is the target distribution, so as to realize the sampling of the target distribution. In this embodiment, the algorithm is used to find the maximum a posteriori probability path in the Bayesian network. Let the path be , and the posterior probability be (where is the observation data), and by continuously iterating the sampling, the path that maximizes is found, and the anti-interference routing strategy is generated.

[0081] In addition, the network topology optimization module also constructs a real-time topology update mechanism. The sliding time window is used to capture the transient change characteristics of the power grid topology, and the length of the sliding time window is set to . In each time window, the change of the power grid topology is monitored and analyzed. At the same time, a prediction correction algorithm based on Kalman filtering is defined to dynamically correct the estimated value of the path delay in the communication routing table. Let the estimated value of the path delay be , and the actual measured value be , according to the Kalman filtering formula (where is the Kalman gain, The correction is made for each time step. A tabu search strategy is used to quickly locate a suboptimal routing solution within a preset solution space. When the main route fails or is severely disturbed, the suboptimal route can be quickly switched to, a dynamic path switching instruction against sudden disturbances is generated, and the continuity of communication is ensured.

[0082] Embodiment 4:

[0083] This embodiment elaborates on the implementation process of the dynamic resource scheduling module. In the actual operation of the railway operation and maintenance communication network, the reasonable allocation of communication resources is the key link to ensure the communication effect and the overall performance of the system, and the dynamic resource scheduling module is the core component to achieve this goal.

[0084] In the communication network along the railway, multiple base stations are distributed, and the communication demand and power grid load of different areas are changing all the time. In order to realize the joint optimization and scheduling of spectrum and power, an effective algorithm mechanism needs to be constructed.

[0085] First, an upper model is established, which optimizes the communication frequency band selection strategy with the goal of minimizing the harmonic interference of the power grid. In practical applications, there are multiple options for communication frequency bands, denoted as set . Different frequency bands are affected differently by the harmonic interference of the power grid. Let the harmonic interference value of frequency band be . Through analysis of historical data and real-time monitoring, the interference values of different frequency bands under different power grid operating conditions are obtained. When selecting a communication frequency band, these interference values are considered comprehensively, and a frequency band with a smaller interference value is selected as much as possible to reduce the impact of harmonic interference of the power grid on communication and ensure communication quality. For example, after comparing the sizes of , , , , if is the smallest, then the frequency band is selected as the priority under the current conditions.

[0086] Next, a lower model is constructed, which solves the power allocation scheme with the constraint of balancing the energy consumption of base stations. Each base station consumes a certain amount of energy when transmitting signals, and different power allocation methods will result in different energy consumption of base stations. Let the set of base stations be , and the power of base station be . The energy consumption of a base station is related to its power, and it is assumed that the energy consumption of a base station is linearly related to its power , where is the energy consumption of base station energy consumption coefficient, which is determined by the characteristics of the base station equipment). To ensure the balance of the energy consumption of the base stations in the entire communication network and avoid premature damage or performance degradation of some base stations due to excessive energy consumption, the upper limit of the total energy consumption is set as , that is, the following condition needs to be met . Under the condition of meeting the constraint, the power is reasonably allocated according to the communication demand of the coverage area of different base stations. For example, the power allocation of the base station in the area with a larger communication demand is appropriately increased, and the power of the base station in the area with a smaller communication demand is correspondingly reduced, so as to ensure the communication quality while achieving the balance of the energy consumption of the base stations.

[0087] The alternating direction multiplier method is used to iteratively solve the double-layer model. This method decomposes the complex double-layer optimization problem into relatively simple sub-problems for solving. In the iteration process, first, the power allocation scheme of the lower layer model is fixed, and the communication frequency band is optimized according to the objective of the upper layer model; then, the communication frequency band selected by the upper layer model is fixed, and the power allocation scheme is adjusted according to the constraint of the lower layer model. This is repeated iteratively to continuously optimize the communication frequency band and power allocation. Each iteration will make the communication frequency band selection more conducive to reducing power grid harmonic interference, and the power allocation more balanced in energy consumption. After multiple iterations, the final spectrum-power joint configuration scheme is output, realizing efficient utilization of communication resources and improving the overall performance of the railway operation communication network.

[0088] Embodiment 5:

[0089] This embodiment mainly introduces the establishment of the abnormality cooperative processing module and the implementation of other functions of the system. In the abnormality cooperative processing module, a lightweight fault diagnosis model is trained at the local node. The transient characteristic data and communication packet loss mode data of the power grid collected locally are trained using a machine learning algorithm. Let the power grid transient characteristic vector be , and the communication packet loss mode vector be , the hidden variable relationship model between the two is established through training.

[0090] A model aggregation protocol is designed to fuse the gradient updates of each node through a differential privacy protection mechanism. Under the distributed federated learning framework, each local node uploads the gradient updates obtained by training to the central server. To protect data privacy, a differential privacy protection mechanism is used. Let the gradient update of node be , and add noise ( subject to a specific noise distribution) before uploading, that is, the uploaded gradient update is . The central server fuses the gradient updates uploaded by each node according to the model aggregation protocol to update the global model.

[0091] A collaborative repair decision tree containing fault isolation instructions and channel switching strategies is generated. According to the trained model and the fused gradient update quantity, when the power grid fault feature or the communication interruption mode is detected, the corresponding collaborative repair decision tree is generated. For example, when a certain regional power grid fault causes communication interruption, the decision tree outputs fault isolation instructions to isolate the fault region, and generates a channel switching strategy to switch the communication to a backup channel, ensuring the normal operation of the communication.

[0092] In addition, the system also constructs a spatio-temporal correlation database, and uses a tensor decomposition algorithm to mine cross-domain association rules between historical power grid events and communication failures. Let the historical power grid event data tensor be , the communication failure data tensor be , and the association rules between the two be mined by a tensor decomposition algorithm (where , , are low-dimensional tensors decomposed from the tensor, is the rank of the decomposition, and is the tensor outer product operation). A rule engine is defined to match the current data pattern in real time, and when the current data pattern matches the mined association rules, a preventive maintenance suggestion queue is generated to take measures in advance to avoid the occurrence of failures.

[0093] The system also deploys an edge computing node cluster to realize end-to-end delay compression of data collection and decision generation using a stream computing framework. The edge computing nodes are distributed along the railway line and can collect data in real time and perform preliminary processing. A stream computing framework such as Apache Flink is used to process the collected data in real time and quickly generate decisions. A priority scheduling algorithm is defined to dynamically allocate computing resources according to the urgency of optimization tasks. Let the set of optimization tasks be , and the urgency of each task be ( ). The tasks are sorted according to the urgency, and computing resources are allocated to handle urgent tasks first to improve the response speed and performance of the system.

[0094] It should be noted that in this paper, relational terms such as first and second are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "contain" or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed or inherent to such a process, method, article or device.

[0095] While embodiments of the application have been shown and described, it is to be understood that the embodiments described are merely exemplary of the principles and application of the present application. Numerous modifications and adaptions can be effected without departing from the spirit and scope of the present application, which is not limited to the exact construction and arrangement described. It is intended, therefore, to cover all modifications and adaptions that fall within the scope of the claims and their equivalents.

Claims

1. A railway operations communication network optimization system that fuses power grid information, characterized by, The system comprises: a multi-source data fusion module for synchronously collecting power load fluctuations, power grid topology states, rail vibration frequency spectra, communication signal strengths and environmental meteorological data through heterogeneous sensors, and generating a time-space multi-dimensional data cube; a communication quality evaluation module for performing nonlinear correlation analysis on the time-space multi-dimensional data cube by using a fuzzy logic fusion algorithm, and generating a communication link quality degradation index sequence; a network topology optimization module for constructing a path planning model based on a dynamic Bayesian network, and calculating an optimal communication routing table under power interference conditions according to the degradation index sequence; a dynamic resource scheduling module for designing a resource allocation algorithm based on double-layer planning, and generating a joint spectrum and power scheduling strategy in combination with real-time power grid loads and communication demand prediction values; an abnormality cooperative processing module for establishing a distributed federated learning framework, and generating a cross-domain cooperative repair instruction set according to power grid fault features and communication interruption modes.

2. The railroad operations communication network optimization system of claim 1, wherein, The implementation of the multi-source data fusion module comprises: spatio-temporal stamp alignment processing of heterogeneous data streams, elimination of sensor sampling rate differences through a sliding window mechanism; construction of a fuzzy logic rule base, mapping of power grid harmonic distortion rates and communication bit error rates into membership vectors; evidence theory synthesis algorithm for credibility weighting of conflicting data, and output of a reduced dimension fusion feature matrix.

3. The railroad operations communication network optimization system of claim 1, wherein, The construction of the communication quality evaluation module comprises: definition of input variables of a fuzzy inference system as signal attenuation gradients, multipath time delay spread and power grid harmonic injection energy; definition of a Gaussian type membership function to depict the degradation trend of each variable, and generation of a three-dimensional rule surface; extraction of key degradation modes through a lambda-cut method, and output of communication link quality level classification results.

4. The railroad operations communication network optimization system of claim 1, wherein, The design of the network topology optimization module comprises: mapping of transformer substation nodes and communication relay stations into a double-layer node structure of the Bayesian network; calculation of a conditional probability distribution table of power grid flow mutation events on communication bit error rates, solution of a maximum a posteriori probability path by using a Markov chain Monte Carlo sampling algorithm, and generation of an anti-interference routing strategy.

5. The railroad operations communication network optimization system of claim 1, wherein, The implementation of the dynamic resource scheduling module comprises: establishment of an upper layer model to minimize power grid harmonic interference as an objective, optimization of a communication frequency band selection strategy, construction of a lower layer model to balance base station energy consumption as a constraint, and solution of a power allocation Pareto frontier; iterative solution of the double-layer model by using an alternating direction multiplier method, and output of a spectrum-power joint configuration scheme.

6. The railroad operations communication network optimization system of claim 1, wherein, The establishment of the abnormality cooperative processing module comprises: training of a lightweight fault diagnosis model at a local node to extract hidden variable relationships between power grid transient features and communication packet loss modes; design of a model aggregation protocol to fuse gradient updates of each node through a differential privacy protection mechanism; generation of a cooperative repair decision tree containing fault isolation instructions and channel switching strategies.

7. The railroad operations communication network optimization system of claim 4, wherein, The network topology optimization module further comprises: construction of a real-time topology update mechanism to capture transient change features of the power grid topology through a sliding time window; definition of a prediction correction algorithm based on Kalman filtering to dynamically correct path delay estimation values in the communication routing table, use of a tabu search strategy to quickly locate a suboptimal routing scheme in a preset solution space, and generation of dynamic path switching instructions against sudden interference.

8. The railroad operations communication network optimization system of claim 1, wherein, The system further comprises: A spatio-temporal database is constructed, and a tensor decomposition algorithm is used to mine cross-domain association rules between historical power grid events and communication failures. A rule engine is defined to match current data patterns in real time and generate a preventive maintenance suggestion queue.

9. The railroad operations communication network optimization system of claim 1, wherein, The system further includes: An edge computing node cluster is deployed, and a stream computing framework is used to realize end-to-end latency compression of data collection and decision generation; A priority scheduling algorithm is defined to dynamically allocate computing resources to handle optimization tasks with different degrees of urgency.

10. A method of optimizing a railway operations communication network with fusion of power grid information, characterized in that, The method includes the following steps: Step 1: Synchronously collect power load fluctuations, power grid topology states, rail vibration spectra, communication signal strengths, and environmental meteorological data through heterogeneous sensors to generate a spatio-temporal multi-dimensional data cube; Step 2: Perform nonlinear correlation analysis on the spatio-temporal multi-dimensional data cube using a fuzzy logic fusion algorithm to generate a communication link quality degradation index sequence; Step 3: Construct a path planning model based on a dynamic Bayesian network, and calculate an optimal communication routing table under power interference conditions according to the degradation index sequence; Step 4: Design a resource allocation algorithm based on double-layer planning, and generate a joint scheduling strategy for spectrum and power in combination with real-time power grid load and communication demand prediction values; Step 5: Establish a distributed federated learning framework, and generate a cross-domain collaborative repair instruction set according to power grid failure characteristics and communication interruption patterns.

Images