Station platform wireless signal coverage optimization method and system based on intelligent algorithm

Abstract

The invention discloses a station platform wireless signal coverage optimization method and system based on an intelligent algorithm, particularly relates to the technical field of mobile communication network coverage optimization, and is used for solving the problems of strategy mutual exclusion and overall performance reduction caused by independent optimization of different systems when a multi-system network shares antenna resources in the prior art. A multi-band antenna unit is deployed, wireless signals and environment data are collected in real time, independent optimization directions of all networks are simulated to predict resource allocation conflicts, strategy exclusiveness is evaluated, a collaboration index is generated, a cooperative game mechanism or a master-slave game mechanism is dynamically selected based on the index to generate a global optimization decision, and the overall optimization decision is obtained. And finally, executing the decision and collecting feedback data to form closed-loop optimization so as to realize intelligent collaboration and dynamic optimization of multi-system network coverage.

Description

Technical Field

[0001] This invention relates to the field of mobile communication network coverage optimization technology, and more specifically, to a method and system for optimizing wireless signal coverage on railway platforms based on intelligent algorithms. Background Technology

[0002] In the platform areas of large transportation hubs such as train stations and subway stations, network operators have generally deployed wireless communication systems covering multiple generations of mobile communication network standards to meet the public's mobile communication needs. Existing technologies typically employ the method of fixing multi-band antenna elements onto the platform structure to provide signal coverage for different network standards such as 2G, 3G, 4G, and 5G. For network optimization, existing solutions mainly rely on independent parameter adjustments and optimizations for each generation of network or its specific frequency band. For example, based on the signal quality measurement reports of each network, the transmit power or mechanical tilt angle of the corresponding antenna element is adjusted independently to achieve the network's own coverage goals and performance indicators.

[0003] However, due to the inherent differences in technical characteristics and service requirements of different mobile communication networks, such as varying requirements for coverage continuity, network capacity, and latency, the aforementioned method based on independent optimization for each network standard can lead to fundamental resource allocation conflicts when they share antenna radiating elements in the same physical space or are deployed in close co-location. Optimization actions taken to prioritize the performance indicators of a certain network standard often damage the performance of other networks, resulting in strategic mutual exclusion between multiple networks. This makes it difficult for the entire communication system to coordinate and efficiently utilize limited wireless resources and antenna degrees of freedom in a dynamically changing environment, thereby causing a decrease in overall system energy efficiency and a loss of comprehensive service quality. Summary of the Invention

[0004] In order to overcome the above-mentioned defects of the prior art, the present invention provides a method and system for optimizing wireless signal coverage on railway platforms based on intelligent algorithms to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: The method for optimizing wireless signal coverage on station platforms based on intelligent algorithms includes: S1. Install several hanging poles under the station platform, and install a multi-band antenna unit supporting multiple network standards on the top of each hanging pole. S2. Real-time acquisition of wireless signal parameters and environmental monitoring data of various network standards fed back by multi-band antenna units; S3. Based on wireless signal parameters and environmental monitoring data, simulate the policy gradient update direction of each network type as an independent optimization agent, and calculate the divergence between all policy gradient update direction vectors to generate a predictive resource allocation conflict degree index. S4. Evaluate the policy mutual exclusion between real-time optimization objective functions of different network types in a multi-network based on the predictive resource allocation conflict degree index, and generate a policy synergy index. S5. Based on the comparison results between the strategy synergy index and the preset threshold, selectively invoke the cooperative game mechanism or the master-slave game mechanism to generate optimization decisions for the overall radiation parameters of the multi-band antenna unit. S6. Execute optimization decisions, dynamically adjust the overall radiation parameters of the multi-band antenna unit, and use the adjusted wireless signal parameters as feedback data.

[0006] Furthermore, S1 includes: The multi-band antenna unit is configured to simultaneously transmit and receive wireless signals belonging to at least two different mobile communication network standards operating frequency bands; Furthermore, it integrates an electronically adjustable drive mechanism into the multi-band antenna unit, enabling independent and continuous adjustment of the overall radiation parameters.

[0007] Furthermore, S2 includes: The monitoring interface of the multi-band antenna unit is used to collect the reference signal received power and signal-to-noise ratio of each network type in the current coverage area as wireless signal parameters in parallel. Meanwhile, environmental monitoring data characterizing passenger flow density and train arrival and departure status are collected by sensors deployed in the station platform environment.

[0008] Furthermore, S3 includes: Using real-time collected wireless signal parameters and environmental monitoring data of each network standard as input, the policy gradient update direction of each network standard at the current moment is determined with the objective function of improving its own network coverage quality. The policy gradient update direction of each network type is represented as a direction vector in a multidimensional parameter space, and the cosine similarity between each pair of direction vectors is calculated. All cosine similarity results are aggregated to generate a scalar value representing the overall degree of strategy dispersion, which serves as an indicator of the degree of predictive resource allocation conflict.

[0009] Furthermore, S4 includes: Based on the predictive resource allocation conflict degree index, the priority direction of the optimization objective function of each network system in the parameter adjustment space is analyzed. Calculate the ratio of the area of ​​the overlapping region of the preferred direction in the multidimensional parameter space to the area of ​​the exclusive region of each direction; The ratio is normalized and mapped to a continuous value in the range of 0 to 1, which serves as a policy synergy index characterizing the degree of policy reconciliation between networks.

[0010] Furthermore, based on the predictive resource allocation conflict degree index, the priority direction of the optimization objective function of each network type in the parameter adjustment space is analyzed, including: based on the divergence degree of the strategy gradient vector reflected by the predictive resource allocation conflict degree index, the most urgent parameter dimension and its adjustment trend that each network type needs to adjust in the multidimensional parameter space to improve its own network coverage quality are inversely inferred, so as to determine their respective priority directions.

[0011] Furthermore, S5 includes: When the strategy synergy index is higher than the preset threshold, a multi-objective optimization problem is constructed using the real-time optimization objective function of each network type. The Pareto optimal solution set is searched in the feasible solution space of the overall radiation parameters through a negotiation game algorithm. An equilibrium solution is selected from the Pareto optimal solution set as the final combination of overall radiation parameters to generate an optimization decision. When the strategy synergy index is not higher than the preset threshold, a dominant network is specified according to the preset priority rules, the optimization objective function of the dominant network is set as the main optimization objective, and the overall radiation parameter adjustment range of other subordinate networks is constrained within the allowable range. Single objective constraint optimization is performed, and the solution obtained is used as the forced synergy scheme and an optimization decision is generated.

[0012] Furthermore, the overall radiation parameters include azimuth angle, downtilt angle, and output power.

[0013] Furthermore, S6 includes: The overall radiation parameter values ​​determined in the optimization decision are converted into a sequence of control commands for the electronically adjustable drive mechanism inside the multi-band antenna element, so as to drive the multi-band antenna element to complete the adjustment of the azimuth angle, downtilt angle and output power; After the adjustment is completed, the wireless signal parameters of each network standard fed back by the multi-band antenna unit are collected again, and they are packaged together with the environmental monitoring data at the corresponding time into a new data record, which is used as feedback data for the generation of predictive resource allocation conflict degree index in subsequent iterations.

[0014] On the other hand, the present invention provides a station platform wireless signal coverage optimization system based on intelligent algorithms, comprising: The unit installation module is used to deploy several hanging poles under the station platform and install a multi-band antenna unit supporting multiple network standards on the top of each hanging pole. The data acquisition module is used to collect wireless signal parameters and environmental monitoring data of various network standards fed back by the multi-band antenna unit in real time. The index generation module is used to simulate the policy gradient update direction of various network standards as independent optimization agents based on wireless signal parameters and environmental monitoring data, and to calculate the divergence between all policy gradient update direction vectors to generate a predictive resource allocation conflict degree index. The mutual exclusion evaluation module is used to evaluate the policy mutual exclusion between real-time optimization objective functions of different network types in a multi-network based on a predictive resource allocation conflict degree index, and to generate a policy synergy index. The decision generation module is used to selectively invoke cooperative game mechanism or master-slave game mechanism based on the comparison result of strategy synergy index and preset threshold to generate optimization decisions for the overall radiation parameters of multi-band antenna unit. The decision execution module is used to execute optimization decisions, dynamically adjust the overall radiation parameters of the multi-band antenna unit, and use the adjusted wireless signal parameters as feedback data.

[0015] Compared with the prior art, the present invention has the following beneficial effects: 1. By treating a multi-standard network as a system requiring overall tuning, it can proactively quantify and assess the potential conflict between different network optimization objectives and dynamically select the most suitable collaborative strategy accordingly. This ensures that every antenna parameter adjustment made for the complex dynamic environment of the platform is the result of weighing the interests of each network from a global system perspective. This fundamentally avoids a zero-sum game situation where the optimization of a single network damages the performance of other networks. It realizes intelligent, dynamic, and efficient allocation of wireless resources and antenna degrees of freedom among multiple networks, significantly improving the overall service capability and energy efficiency of the system under limited physical resources.

[0016] 2. By simulating the independent optimization tendencies of each network and predicting conflicts, the mutual exclusivity of strategies is evaluated and a clear synergy index is generated. Finally, based on this index, cooperative or master-slave game mechanisms are adaptively invoked to generate optimization decisions. This transforms the overall radiation parameter adjustment of multi-band antenna elements from a passive response to changes in the quality of a single network into an intelligent process that can anticipate conflicts, actively coordinate, and seek global Pareto improvement or safeguard core network objectives. This not only improves the accuracy and synergy of network coverage optimization in multi-standard coexistence scenarios, but also enhances the system's adaptability to dynamic environmental changes such as platform passenger flow and train arrivals and departures, thereby providing better overall coverage performance in complex station wireless environments. Attached Figure Description

[0017] Figure 1 This is a flowchart of the intelligent algorithm-based method for optimizing wireless signal coverage on railway platforms according to the present invention. Figure 2 This is a schematic diagram of the structure of the station platform wireless signal coverage optimization system based on intelligent algorithms according to the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] Example 1: Figure 1 This invention presents a method for optimizing wireless signal coverage on railway platform based on intelligent algorithms, comprising: S1. Install several hanging poles under the station platform, and install a multi-band antenna unit supporting multiple network standards on the top of each hanging pole. S2. Real-time acquisition of wireless signal parameters and environmental monitoring data of various network standards fed back by multi-band antenna units; S3. Based on wireless signal parameters and environmental monitoring data, simulate the policy gradient update direction of each network type as an independent optimization agent, and calculate the divergence between all policy gradient update direction vectors to generate a predictive resource allocation conflict degree index. S4. Evaluate the policy mutual exclusion between real-time optimization objective functions of different network types in a multi-network based on the predictive resource allocation conflict degree index, and generate a policy synergy index. S5. Based on the comparison results between the strategy synergy index and the preset threshold, selectively invoke the cooperative game mechanism or the master-slave game mechanism to generate optimization decisions for the overall radiation parameters of the multi-band antenna unit. S6. Execute optimization decisions, dynamically adjust the overall radiation parameters of the multi-band antenna unit, and use the adjusted wireless signal parameters as feedback data.

[0020] S1. Install several mounting poles under the station platform, and install a multi-band antenna unit supporting multiple network standards at the top of each mounting pole. The specific implementation is as follows: Several mounting poles are deployed beneath the station platform, with a multi-band antenna unit supporting multiple network standards installed at the top of each pole. These multiple network standards refer to currently used mobile communication networks of different generations, such as a combination of at least second-generation and fifth-generation mobile communication networks, or a combination of third-generation and fourth-generation mobile communication networks, or any other combination of two or more network standards. When installing the multi-band antenna unit, it is physically fixed to the top of the mounting pole and connected to the baseband processing unit and radio frequency remote unit located in the equipment room via feed lines, forming a complete signal transmission and reception link. The location and number of mounting poles are planned based on the specific length and structural layout of the station platform, as well as the passenger waiting area to be covered, with the basic goal of ensuring no coverage blind spots.

[0021] A multi-band antenna unit is configured to simultaneously transmit and receive radio signals belonging to at least two different mobile communication network standards operating frequency bands. This configuration is implemented in hardware through multiple sets of radiating elements integrated within the multi-band antenna unit, along with corresponding filters and combiners. For example, a multi-band antenna unit may integrate a first set of radiating elements, whose resonant frequency is designed and tuned in the 700 MHz to 900 MHz frequency band to cover the low-frequency operating frequencies of second-generation (G2) and some fourth-generation (4G) mobile communication networks. Simultaneously, the antenna unit integrates a second set of radiating elements, whose resonant frequency is designed and tuned in the 3.3 GHz to 3.6 GHz frequency band to cover the specific operating frequencies of fifth-generation (5G) mobile communication networks. The two sets of radiating elements combine and separate signals through an internal frequency combiner, allowing multiple radio frequency signals of different frequencies from the base station equipment to be transmitted to the antenna through the same feeder and simultaneously radiated outwards by the corresponding radiating elements. Conversely, wireless signals received by the antenna from different frequency bands can be separated and transmitted to their respective network receivers via the same path. This hardware configuration enables a single physical antenna element to simultaneously serve multiple networks of different standards, achieving physical-level co-location and resource sharing.

[0022] Furthermore, the multi-band antenna unit integrates an electrically adjustable drive mechanism capable of independently and continuously adjusting the overall radiation parameters. These overall radiation parameters specifically include the antenna's pointing angle (azimuth angle) in the horizontal direction, the tilt angle (downtilt angle) of the main beam in the vertical direction, and the power level of the transmitted radio frequency signal (output power). Physically, the electrically adjustable drive mechanism is an electromechanical integrated component within the multi-band antenna unit housing. It typically includes a horizontal stepper motor and transmission gear set for controlling horizontal rotation, a vertical stepper motor and worm gear mechanism for controlling vertical pitch, and a digitally controlled electrically adjustable attenuator circuit for controlling the attenuation of the radio frequency signal. The horizontal stepper motor receives external control signals and, through the gear set, drives the antenna radiator to rotate horizontally within an azimuth angle range of 0 to 360 degrees, with a rotational step accuracy of 0.1 degrees, thereby achieving continuous and precise adjustment of the azimuth angle. The vertical stepper motor also receives external control signals and drives the antenna radiator to tilt vertically within a range of 0 to 15 degrees via a worm gear mechanism. The tilting step accuracy can also reach 0.1 degrees, thus achieving continuous and precise adjustment of the downtilt angle. The digitally controlled electrically adjustable attenuator circuit is integrated into the RF signal path. Based on the received external voltage or digital control signal, it continuously adjusts the attenuation of the RF signal within a range of 0 to 30 dB, thereby effectively changing the output power of the antenna port with an adjustment accuracy of 0.5 dB.

[0023] All electrically adjustable drive mechanisms are connected to a central controller via a standardized control interface. Based on optimization decisions generated in subsequent optimization steps, the central controller converts desired parameter values—such as a azimuth angle of 120.5 degrees, a downtilt angle of 8.2 degrees, and an output power of 18 dBmW—into a series of drive level signals or pulse sequences. These signals are then sent to the corresponding motors and circuits within the electrically adjustable drive mechanisms via the control interface, driving them to complete mechanical movements or electrical parameter changes. Ultimately, this ensures that the antenna's actual radiation state matches the requirements of the optimization decisions. This integration of the electrically adjustable drive mechanisms eliminates the need for manual on-site climbing operations, enabling remote, automatic, and rapid adjustments to antenna coverage performance. This provides the necessary physical execution capability for subsequent dynamic optimization.

[0024] S2. Real-time acquisition of wireless signal parameters and environmental monitoring data of various network standards fed back by multi-band antenna units, specifically implemented as follows: The monitoring interface of the multi-band antenna unit collects the reference signal received power and signal-to-interference-plus-noise ratio (SNR) of each network standard in the current coverage area in parallel as wireless signal parameters. Specifically, the multi-band antenna unit installed in step S1 or its connected remote radio unit is designed with a standardized monitoring interface that can continuously output physical layer measurement reports generated by the baseband chips of each network standard. For a multi-band antenna unit supporting at least two network standards, its internal processor or connected processor independently demodulates and analyzes the signal of each network standard. For second-generation mobile communication networks, the collected parameter is the received signal strength indicator, whose physical meaning is similar to the reference signal received power, characterizing the signal field strength. For third-generation, fourth-generation, and fifth-generation mobile communication networks, the reference signal received power defined by the network standard is directly collected. The collection process is as follows: for example, every 100 milliseconds is a measurement cycle. In each cycle, the monitoring interface synchronously outputs the reference signal received power values ​​of all active networks standard. This value is a continuous variable, typically ranging from -120 dBmW to -50 dBmW. The SNR is collected synchronously. The processing logic within the monitoring interface measures the ratio of the sum of the received useful signal power and the interference noise power within the operating frequency band of each network standard, and converts this ratio to decibels (dB) through logarithmic calculation. For example, for fourth-generation mobile communication networks, the signal-to-interference-noise ratio (SNR) is calculated based on measurements of the cell-specific reference signal, with a typical range between -3 dB and 30 dB. Both parameters, when output from the monitoring interface, are marked with a corresponding timestamp and network standard identifier, such as 5G_NR_RSRP or 4G_LTE_SINR, thus ensuring clear data attribution and synchronization.

[0025] Simultaneously, environmental monitoring data characterizing passenger flow density and train arrival / departure status are collected by sensors deployed in the station platform environment. This requires establishing a sensor information network independent of communication equipment. For passenger flow density, implementation methods include, but are not limited to, installing visual sensors or infrared beam sensors with people counting functions at key locations such as platform entrances, the middle and rear of the waiting area. The visual sensors, through their built-in edge computing units, perform real-time analysis of captured video images, using background subtraction and contour recognition methods to count the number of passengers in a defined area, and then divide this number by the area to obtain the passenger flow density value per unit area, for example, 0.2 people per square meter. Passenger flow density data is updated at a frequency of, for example, once per second, and is also appended with a timestamp and sensor location number. For train arrival / departure status, implementation methods include installing axle counting sensors on the trackside or interfacing with the station's train control system. Axle counting sensors detect the passing of train wheels using magnetic induction principles, thereby determining the train's entry or departure from the platform section. A more direct method is to obtain real-time information on train schedules and actual arrival / departure times from the train control system. Regardless of the specific sensor or interface used, the final environmental monitoring data contains at least two logical states: train stopped or train not stopped, and can be expanded to include the specific platform number where the train stopped. This status information is updated each time a train arrives or departs. All raw passenger flow density and train arrival / departure status data acquired from different sensors are sent to a unified data aggregation point, synchronized with the aforementioned wireless signal parameters, and together constitute a real-time data record containing multi-dimensional fields.

[0026] S3. Based on wireless signal parameters and environmental monitoring data, simulate the policy gradient update direction of various network standards as independent optimization agents, and calculate the divergence between all policy gradient update direction vectors to generate a predictive resource allocation conflict degree index. The specific implementation is as follows: Using real-time collected wireless signal parameters and environmental monitoring data for each network type as input, the policy gradient update direction for each network type at the current moment is determined, with the objective function being to improve its own network coverage quality. First, an instantaneous optimization objective function is defined for each network type. This function is a mathematical expression whose output value characterizes the coverage quality of the network at the current location; a larger value generally indicates better coverage quality. The input variables of the function are the adjustable overall radiation parameters defined in step S1, namely, azimuth angle, downtilt angle, and output power. The specific composition of the function is determined by the wireless signal parameters and environmental monitoring data at the current moment. For example, for a certain fourth-generation mobile communication network, its current optimization objective function can be constructed as follows: using the average reference signal received power value of all sampling points within the current coverage area of ​​the network as a base term, subtracting the proportion of sampling points with reference signal received power below a preset threshold as a penalty term, and adding a fixed reward term when the current train arrival / departure state is train stop. In this context, the average reference signal received power is directly derived from the measurement value in step S2; the proportion below the threshold is obtained by statistically analyzing the measurement values; and the train arrival / departure status reward directly uses the environmental monitoring data from step S2. The preset threshold mentioned here is set based on the coverage quality specifications of mobile communication networks. This threshold is typically determined based on the minimum signal strength required for the network to provide basic services, such as voice calls. For example, for fourth-generation mobile communication networks, the typical reference signal received power that ensures the continuity of basic services for edge users is approximately -105 dBmW to -110 dBmW. Therefore, in this embodiment, -105 dBmW can be selected as the preset threshold. During function calculation, the real-time measurement values ​​of all reference signal received power collected in step S2 are compared one by one with the preset threshold, and the percentage of sample points with measurement values ​​less than the preset threshold is calculated out of the total number of sample points. This percentage is the value of the penalty item. This function explicitly expresses the network's optimization intention under the current environment: to improve average signal strength, reduce the proportion of weak coverage points, and achieve higher quality evaluation when trains arrive at stations with concentrated passenger traffic. After defining this optimization objective function, the gradient of this function with respect to the three-dimensional variables of the overall radiation parameters is calculated. The gradient is a vector, where each component is the value obtained by taking the partial derivative of the objective function with respect to the azimuth angle, downtilt angle, and output power, respectively. The direction of this gradient, that is, the direction in which the function value increases the fastest in the three-dimensional parameter space consisting of the azimuth angle, downtilt angle, and output power, is determined as the policy gradient update direction of the network at the current moment. It indicates how the network should fine-tune its antenna parameters to maximize the improvement of coverage quality if it only considers its own interests.

[0027] The policy gradient update direction of each network type is represented as a direction vector in a multidimensional parameter space, and the cosine similarity between each pair of direction vectors is calculated. Following the previous step, the policy gradient update direction of each network type is mathematically expressed as a three-dimensional vector. For example, vector A equals the value a1 for the direction angle change component, a2 for the downtilt angle change component, and a3 for the output power change component. To eliminate the influence of potential absolute numerical differences in the dimensions of the objective function or computational scale between different network gradient vectors on subsequent comparisons, each gradient vector needs to be normalized to a length of 1, retaining only the direction information. After normalization, the optimization intention of each network type is purified into a unit direction vector. Subsequently, the cosine similarity between each pair of unit direction vectors of all networks is calculated. For any two network unit direction vectors, such as vector B and vector C, the cosine similarity is equal to the dot product of vectors B and C. The dot product is calculated by multiplying the corresponding components of the two vectors in each dimension and then summing the results. Since they are unit vectors, the result of the dot product is directly equal to the cosine of the angle between the two vectors, with a value ranging from -1 to +1. When the two vectors are in the same direction, the cosine similarity is +1, indicating that the independent optimization strategies of the two networks are completely consistent and there is no conflict. When the two vectors are perpendicular, the cosine similarity is 0. When the two vectors are in opposite directions, the cosine similarity is -1, indicating the most severe policy conflict. By calculating the cosine similarity between all network pairs—for example, when there are three networks, three sets of cosine similarity values ​​need to be calculated: network pair 1 (network 2), network pair 2 (network 3), and network pair 2 (network 3)—to obtain a similarity set that characterizes the consistency of policy direction between any two networks.

[0028] All cosine similarity results are aggregated to generate a scalar value representing the overall policy dispersion level, serving as a predictive indicator of resource allocation conflict. After obtaining pairwise cosine similarity sets, they need to be merged into a single quantitative indicator that can reflect the overall dispersion level of all network policies from a global perspective. A direct and effective aggregation method is to calculate the arithmetic mean of all calculated cosine similarities. However, a simple arithmetic mean may mask extreme conflict situations. Therefore, a more robust aggregation logic is to first set an ideal policy consistency threshold, such as 0.8. The basis for setting this ideal policy consistency threshold is that a cosine similarity of 1 represents complete consistency, but in actual multi-parameter joint optimization, due to the complexity of the optimization problem, the gradient directions of each network policy are completely consistent, i.e., a similarity of 1 is an ideal state, which is difficult to achieve. It is generally believed that when the cosine similarity is greater than 0.8, the optimization directions of the two networks can be considered highly coordinated, and the risk of resource conflict is very low. Therefore, 0.8 is selected as the ideal policy consistency threshold to measure the gap between the actual similarity and the highly coordinated state. Then, each cosine similarity value is compared with this ideal policy consistency threshold. For each network pair, the difference between its similarity value and 0.8 is calculated, and the absolute value is taken. The magnitude of this absolute difference intuitively reflects the degree to which the policy direction of the network pair deviates from the ideal highly cooperative state. Finally, these absolute differences for all network pairs are summed, and the sum is divided by the total number of network pairs to obtain an average deviation. The average deviation is a scalar value; the smaller the value, the more concentrated and consistent the policy directions of all networks; the larger the value, the more dispersed the policy directions and the more likely there is resource contention. This scalar value obtained through aggregation calculation is defined as the predictive resource allocation conflict degree index. This index integrates the independent optimization tendencies of all network pairs, achieving a quantitative prediction of possible future parameter adjustment conflicts.

[0029] S4. Evaluate the policy mutual exclusion between real-time optimization objective functions of different network standards in a multi-standard network based on the predictive resource allocation conflict degree index, and generate a policy synergy index. The specific implementation is as follows: This paper analyzes the priority direction of the optimization objective function of each network type in the parameter adjustment space based on the predictive resource allocation conflict degree index. Specifically, it is necessary to first understand the essence of the predictive resource allocation conflict degree index. This index is aggregated from the cosine similarity of the policy gradient direction vectors of all networks in step S3. Its value directly reflects the overall divergence of the vector in the multi-dimensional parameter space composed of orientation angle, downtilt angle, and output power. A high degree of divergence means that the parameter dimensions and adjustment trends that each network urgently wants to adjust to improve its own coverage quality are very different. Analyzing the priority direction means inversely inferring the strongest optimization intention of each individual network from this overall divergence state. One implementation method is to review the intermediate data stored in step S3 for calculating this index, that is, the original policy gradient update direction vector of each network type after gradient calculation and before normalization. The three components of this vector represent the instantaneous growth rate of the network coverage quality function with parameter changes in the three dimensions of orientation angle, downtilt angle, and output power, respectively. The absolute value of a component indicates the urgency of adjustment in that dimension; the sign of the component indicates whether the adjustment trend is to increase or decrease the parameter value. For example, for the original policy gradient update direction vector of a certain network, the absolute value of its output power component is significantly larger than the other two components, and its sign is negative. This indicates that, at the current moment, the most sensitive and effective adjustment to improve the network's coverage quality is to reduce the output power, while adjusting the direction angle or downtilt angle has a relatively insignificant effect. Therefore, the preferred direction of this network can be interpreted as: in the parameter adjustment space, optimization is mainly performed along the direction of reducing output power. By analyzing the original policy gradient update direction vector of each network in turn, the preferred direction of each network can be obtained. This direction is jointly defined by the dominant adjustment dimension identifier and the adjustment trend.

[0030] The ratio of the overlapping area of ​​the preferred directions to the area of ​​each direction's exclusive region in the multidimensional parameter space is calculated. After resolving the preferred directions of each network, it is necessary to quantify the conflict and overlap of the optimization intentions represented by the preferred directions in the parameter space. First, an intention region is defined for each network's preferred direction. Since the preferred direction clearly defines the dominant adjustment dimension and trend, such as increasing the direction angle or decreasing the output power, it needs to be transformed into a specific parameter adjustment range in that dimension. The basis and implementation method for setting this range are as follows: For the dimension clearly pointed to by the preferred direction, its adjustment range is defined as a typical adjustment amplitude extending from the current value of the parameter towards the adjustment trend direction. The determination of this typical adjustment amplitude is not a fixed value, but is based on the statistical analysis of the historical adjustment records of the parameter. For example, the absolute value of the actual change value of the parameter after successful optimization in similar network states over a period of time can be statistically analyzed, and the median of the historical change values ​​can be taken as the typical adjustment amplitude at the current moment. If the network does not have a clear preferred direction in that dimension, its intention region in that dimension is defined as a tolerance interval centered on the current parameter value. The radius of this tolerance interval can be set, for example, to five percent of the physically adjustable range of the parameter, to reflect the network's insensitivity to adjustments in this dimension. Following the above method, the intention region of each network constitutes a cuboid in three-dimensional space. Next, the intersection of all network intention region cuboids is calculated; this intersection volume is the overlapping region area, representing the parameter adjustment range that all networks can commonly accept. Then, the volume of each network intention region cuboid is subtracted from the intersection volume to obtain the exclusive region area of ​​that network. The exclusive region areas of all networks are then summed to obtain the total exclusive region area in each direction. Finally, the ratio of the overlapping region area to the sum of the overlapping region area and the exclusive region areas in each direction is calculated.

[0031] The ratio is normalized and mapped to a continuous value between 0 and 1, serving as a strategy synergy index characterizing the degree of policy reconciliation between networks. The area ratio calculated in the previous step is itself a number between 0 and 1, but to make it more sensitive to critical changes in synergy when used as the final decision index, a non-linear normalization mapping is required. This mapping is implemented through a parameter-configurable logic function. The standard form of the logic function requires determining two key parameters: the center point and the growth rate. In this implementation scenario, the center point is set to 0.5, indicating that when the area ratio is 0.5, the mapped strategy synergy index is 0.5. The growth rate needs to be set according to the decision requirements: if it is desired that the index changes smoothly when the ratio is close to 0.5, and changes drastically at both ends, a larger growth rate value is set, such as 10; if a relatively uniform index change is desired, a smaller growth rate value can be set, such as 4. The specific mapping process involves subtracting the center point's 0.5 from the area ratio calculated in the previous step, multiplying by the set growth rate, and using the result as the input variable for a logical function. The function's output value is the final strategy synergy index. By adjusting the growth rate parameter, the sensitivity of this index to the degree of intermediate overlap can be controlled, thus adapting to different optimization decision-making styles. The final generated strategy synergy index is a continuous scalar value within the range of 0 to 1. The closer the value is to 1, the lower the policy mutual exclusion between the real-time optimization objective functions of different network types, and the higher the synergy; the closer the value is to 0, the higher the policy mutual exclusion and the lower the synergy. This index provides a direct and quantitative decision-making basis for step S5 to select a cooperative game mechanism or a master-slave game mechanism.

[0032] S5. Based on the comparison results between the strategy synergy index and the preset threshold, selectively invoke the cooperative game mechanism or the master-slave game mechanism to generate optimization decisions for the overall radiation parameters of the multi-band antenna element. The specific implementation is as follows: The preset threshold is the game mechanism selection threshold. The basis and implementation method for setting the game mechanism selection threshold are as follows: This threshold is obtained through an offline statistical analysis process. Specifically, after the initial deployment of the system and before the automatic optimization during formal operation, a historical data collection phase lasting several weeks is set up. During this phase, the system executes steps S1 to S4 normally, but does not execute the decision in step S5 or the adjustment in step S6. It only records a series of strategy synergy index values ​​generated in various real station platform scenarios, forming a historical dataset. Subsequently, statistical analysis is performed on this historical dataset. Multiple percentiles of the dataset are calculated, such as the 50th percentile, 70th percentile, 80th percentile, etc. The specific values ​​corresponding to the percentiles and the theoretical applicability of different game mechanisms are analyzed, and finally, a value is selected as a fixed preset game mechanism selection threshold. For example, if analysis reveals that when the strategy synergy index value is higher than the 70th percentile of historical data, the conflict between the network objective functions in the corresponding scenario is relatively small, and the success rate and efficiency of cooperative game theory are higher, then the specific value corresponding to the 70th percentile, for example, calculated to be 0.72, can be determined as the preset threshold for selecting the game mechanism. This value is then permanently written into the system's configuration file. In subsequent online operation, each decision is made by directly comparing the real-time generated strategy synergy index with this fixed value read from the configuration file.

[0033] When the strategy synergy index exceeds the preset game mechanism selection threshold, a multi-objective optimization problem is constructed using the real-time optimization objective functions of each network type. A Pareto optimal solution set is searched in the feasible solution space of the overall radiation parameters using a negotiation game algorithm. An equilibrium solution is selected from the Pareto optimal solution set as the final combination of overall radiation parameters, generating an optimization decision. This branch implies significant synergy potential among the network objectives, making it suitable for cooperative game theory through equal negotiation. First, the optimization objective function of each network type at the current moment—the function used to calculate the gradient in step S3, with the orientation angle, downtilt angle, and output power as independent variables—is taken as an independent objective in the multi-objective optimization problem. The feasible solution space of the overall radiation parameters is defined by the physical adjustable range of each parameter, for example, the orientation angle between 0 and 360 degrees, the downtilt angle between 0 and 15 degrees, and the output power between 10 dB / mW and 40 dB / mW. Next, a negotiation game algorithm is applied to solve the problem. One feasible algorithm is a weighted summation method combined with iterative negotiation, as detailed below. The first step is to initialize the weights, assigning an initial weight to the objective function of each network. All weights sum to 1, and the initial weights can be equally distributed. The second step is to perform weighted sum optimization. Within the feasible solution space, find the parameter solution that maximizes the weighted sum of the objective functions of all networks. This is a single-objective optimization problem and can be solved using the gradient ascent method. The third step is to calculate the benefit achievement degree of each network. Substitute the parameter solution obtained in the second step into the objective function of each network to obtain the function value. Then divide this function value by the theoretically optimal function value that the network could achieve under the current conditions when optimized alone, to obtain the benefit achievement degree of each network, which is a number between 0 and 1. The fourth step is to determine the negotiated equilibrium. Calculate the variance of the benefit achievement degrees of all networks. If the variance is less than a minimum value, such as 0.0001, or the number of iterations exceeds a preset limit, such as 20, then proceed to the seventh step to output the final result; otherwise, continue to the fifth step. The fifth step is to dynamically adjust the weights based on the degree of benefit attainment. The network with the lowest benefit attainment has its weight increased by one adjustment step size in the next iteration, for example, 0.05. Simultaneously, the weights of other networks are reduced proportionally to maintain a total weight of 1. The sixth step is to repeat the iteration. Based on the adjusted weights, return to the second step to begin a new round of weighted sum optimization, benefit attainment calculation, and equilibrium judgment. The seventh step is termination and output. When the termination condition of the fourth step is met, the parameter solution that maximizes the weighted sum obtained in the last iteration, which is an equilibrium solution satisfying the Nash bargaining solution, is used as the final overall radiation parameter combination. The specific values ​​of the azimuth angle, downtilt angle, and output power corresponding to this combination constitute the optimization decision generated under this branch.

[0034] When the strategy synergy index is not higher than the preset game mechanism selection threshold, a dominant network is designated according to the preset priority rules. The optimization objective function of the dominant network is set as the main optimization objective, while the overall radiation parameter adjustment range of other subordinate networks is constrained within the allowable range. Single-objective constraint optimization is performed, and the resulting solution is used as the forced synergy scheme to generate an optimization decision. This branch implies a weak synergy foundation, requiring the designation of a dominant network for forced synergy. First, the dominant network is determined according to the preset priority rules. These rules are predefined; for example, the priority order is: fifth-generation mobile communication networks take precedence over fourth-generation mobile communication networks, and fourth-generation mobile communication networks take precedence over third-generation mobile communication networks. Following this order, the currently active networks are checked, and the one ranked highest is designated as the dominant network. Once the dominant network is determined, such as the fifth-generation mobile communication network, its optimization objective function is used as the single maximization objective of the entire optimization problem. For other subordinate networks, the adjustment of their overall radiation parameters is constrained within the allowable range. The method for defining the allowable interval is as follows: for each overall radiation parameter of each subordinate network, using its current value as the center point, extend upwards and downwards by a maximum tolerable variation to form a closed interval. The maximum tolerable variation is determined based on the maximum range of parameter variation that the network can tolerate to maintain basic service quality. For example, through link budget calculations, the maximum tolerable variation in output power can be set to 3 dBmW. Therefore, if the current output power of a subordinate network is 20 dBmW, its allowable interval is 17 dBmW to 23 dBmW. The allowable interval for the dominant network's own parameters is its complete physical adjustable range. Subsequently, under this constraint—that is, the objective is to maximize the objective function of the dominant network, the decision variables are the overall radiation parameters of all networks, and each parameter variable must fall within its own defined allowable interval—a single-objective constrained optimization problem is formed. This problem can be solved using sequential quadratic programming. This method employs an iterative approach. In each iteration, the original constrained optimization problem is approximated as a quadratic programming subproblem near the current solution. A linear search is then performed along the solution direction of the subproblem, updating the current solution until the change in the solution is less than a minimum threshold. The solution obtained at this point is the forced cooperative scheme. The overall radiation parameter combination corresponding to this scheme serves as the optimization decision generated under this branch. Regardless of which branch is used for generation, the optimization decision is ultimately concretized into a set of explicit set values ​​for the three overall radiation parameters of the multi-band antenna element: azimuth angle, downtilt angle, and output power, for execution in step S6.

[0035] S6. Execute optimization decisions, dynamically adjust the overall radiation parameters of the multi-band antenna elements, and use the adjusted wireless signal parameters as feedback data. Specifically, this is implemented as follows: The overall radiation parameter values ​​determined in the optimization decision are converted into a sequence of control commands for the electrically adjustable drive mechanism inside the multi-band antenna unit, thereby driving the multi-band antenna unit to adjust its azimuth, downtilt, and output power. Specifically, the core of the optimization decision generated in step S5 is to define a set of overall radiation parameter values ​​to be set, including a azimuth angle value, a downtilt angle value, and an output power value. The values ​​of the overall radiation parameters need to be accurately transmitted and applied to the multi-band antenna unit with integrated electrically adjustable drive mechanism deployed in step S1. The conversion and driving process is as follows: First, the central controller reads the specific values ​​of azimuth, downtilt, and output power contained in the optimization decision, such as a azimuth angle of 120.5 degrees, a downtilt angle of 8.2 degrees, and an output power of 18 dBmW. The units and dimensions must be completely matched with the design specifications of the electrically adjustable drive mechanism. Next, the central controller converts these target parameter values ​​into a sequence of low-level control commands that the electrically adjustable drive mechanism can recognize and execute, according to preset command encoding rules. The commands are generated based on the physical characteristics and communication protocols of each actuator in the electrically adjustable drive mechanism. For horizontal stepper motors controlling the azimuth angle, drive commands need to be generated. The central controller first queries the currently stored azimuth angle value of the antenna unit and calculates the difference between the target value of 120.5 degrees and the current value. Then, based on the transmission parameters of the motor and gear set—namely, the number of pulses required per revolution of the motor and the gear ratio that converts the motor rotation into an antenna angle change—it calculates the precise total number of pulses required to rotate the antenna to the target angle. For example, if the transmission system parameters are 200 pulses per motor revolution and the gear ratio is 1 degree of antenna rotation for every 100 revolutions of the motor, then rotating 1 degree requires 200 multiplied by 100, which equals 20,000 pulses. Therefore, rotating 0.5 degrees requires 10,000 pulses. The command sequence must include the rotation direction, the total number of pulses, and the pulse frequency. For vertical stepper motors controlling the downtilt angle, the same method is used to generate commands, but their transmission parameters are independent. For CNC ESC attenuator circuits controlling output power, the command is usually a specific digital quantity. The central controller, based on the target output power value, such as 18 dB / mW, queries a pre-stored power setting value-control code mapping table to find the corresponding digital control code, such as the hexadecimal number 0x12. This mapping table is established and stored during antenna unit installation and commissioning through calibration measurements. Then, the central controller encapsulates the instruction according to a preset communication frame format via a standardized control interface, such as an RS-485 bus. A complete control frame may include a start symbol, antenna address code, instruction type, azimuth pulse count, downtilt pulse count, power control code, checksum, and end symbol. This control frame is sent to the target multi-band antenna unit. The control circuitry inside the antenna unit, such as a microcontroller, receives and parses the frame, driving the horizontal stepper motor, vertical stepper motor, and CNC electrically adjustable attenuator circuit according to the instruction content.The system uses feedback from the motor's built-in encoder or Hall sensor to confirm whether the rotation is in place, or detects the RF signal power to confirm whether the output power meets the standard, ensuring that the final physical state of the antenna is consistent with the requirements of the optimization decision. This adjustment process is usually completed within a few seconds.

[0036] After adjustment, the wireless signal parameters of each network standard fed back by the multi-band antenna unit are collected again, and encapsulated together with the corresponding environmental monitoring data into a new data record. This data serves as feedback data for generating predictive resource allocation conflict level indicators in subsequent iterations. Once the antenna adjustment is confirmed, the system immediately initiates a new round of data acquisition to evaluate the actual effect of this optimization decision. The data acquisition fully reuses the mechanism and interface of step S2, but has a clear temporal purpose. The system collects the reference signal received power and signal-to-interference-plus-noise ratio of each network standard under the current new parameter configuration in parallel through the monitoring interface of the multi-band antenna unit. This acquisition process occurs after a brief stabilization period following the adjustment, for example, waiting 3 seconds to allow the wireless channel to stabilize. This 3-second waiting time is empirically set based on the large-scale fading characteristics of wireless signals in the platform environment. Simultaneously, the system collects the passenger flow density and train arrival / departure status at this moment through sensors deployed in the station platform environment. The environmental monitoring data collected at this time characterizes the external environment faced by the network after executing the optimization decision. Next, data encapsulation is performed. The newly collected wireless signal parameters and environmental monitoring data are packaged into a new data record according to a predefined, fixed data format. The encapsulation operation is performed by a data formatting program. This program first obtains a current timestamp accurate to milliseconds. Then, it arranges the following data items sequentially and writes them into a data buffer or database record: a timestamp field; a unique antenna element identifier field; for each network standard, a floating-point value for its reference signal received power and a floating-point value for the signal-to-interference-plus-noise ratio; a floating-point value for passenger flow density; and a Boolean or enumerated value for train arrival and departure status. All values ​​are stored in their physical units, such as decibels per milliwatt, decibels, and people per square meter. The key to encapsulation is ensuring that all data items share the same timestamp and are explicitly associated with the specific antenna element being adjusted. This fully formatted data record is stored as feedback data and added to the historical database. This feedback data serves as key input data for step S3 in the next optimization cycle. Specifically, when step S3 is triggered in the next round, the system program will retrieve the most recent feedback data record from the database—that is, the record stored during the execution of step S6—as the latest sample representing the current state of the network and environment. This sample will then be used to calculate the new policy gradient update direction, thereby generating a new predictive resource allocation conflict level index. In this way, step S6 transforms the results of this optimization into empirical inputs that drive the next optimization, thus forming a complete closed-loop optimization system that continuously learns and dynamically adjusts.

[0037] Example 2: Figure 2A schematic diagram of the station platform wireless signal coverage optimization system based on intelligent algorithms of the present invention is given. The station platform wireless signal coverage optimization system based on intelligent algorithms includes: The unit installation module is used to deploy several hanging poles under the station platform and install a multi-band antenna unit supporting multiple network standards on the top of each hanging pole. The data acquisition module is used to collect wireless signal parameters and environmental monitoring data of various network standards fed back by the multi-band antenna unit in real time. The index generation module is used to simulate the policy gradient update direction of various network standards as independent optimization agents based on wireless signal parameters and environmental monitoring data, and to calculate the divergence between all policy gradient update direction vectors to generate a predictive resource allocation conflict degree index. The mutual exclusion evaluation module is used to evaluate the policy mutual exclusion between real-time optimization objective functions of different network types in a multi-network based on a predictive resource allocation conflict degree index, and to generate a policy synergy index. The decision generation module is used to selectively invoke cooperative game mechanism or master-slave game mechanism based on the comparison result of strategy synergy index and preset threshold to generate optimization decisions for the overall radiation parameters of multi-band antenna unit. The decision execution module is used to execute optimization decisions, dynamically adjust the overall radiation parameters of the multi-band antenna unit, and use the adjusted wireless signal parameters as feedback data.

[0038] All calculations involved in the embodiments are dimensionless numerical calculations, and the preset parameters and thresholds in the calculations are set by those skilled in the art according to the actual situation.

[0039] It should be noted that this invention can be deployed on the device itself to realize embedded applications, or it can run on a PC or other terminal with a user interface, thereby meeting various hardware environments and usage requirements.

[0040] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wireless or wired transmission; wired transmission methods include optical fiber, twisted pair, coaxial cable, etc.; wireless transmission includes infrared, microwave, etc. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center containing one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. A semiconductor medium can be a solid-state drive.

[0041] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and modules described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0042] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or modules may be electrical, mechanical, or other forms.

[0043] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0044] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0045] If the aforementioned functions are implemented as software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0046] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0047] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for optimizing wireless signal coverage of a station platform based on an intelligent algorithm, characterized in that, The method comprises the following steps: S1, arranging a plurality of hanging and holding poles under the station platform, and installing a multi-frequency antenna unit supporting multiple network modes at the top of each hanging and holding pole; S2, collecting the wireless signal parameters and environmental monitoring data of each network mode fed back by the multi-frequency antenna unit in real time; S3, simulating the policy gradient update direction of each network mode when it is an independent optimization agent based on the wireless signal parameters and environmental monitoring data, and calculating the divergence between all policy gradient update direction vectors to generate a predictive resource allocation conflict degree index; S4, evaluating the policy mutual exclusivity between the real-time optimization objective functions of different network modes in the multi-network mode based on the predictive resource allocation conflict degree index, and generating a policy coordination degree index; S5, selectively invoking a cooperative game mechanism or a master-slave game mechanism based on the comparison result of the policy coordination degree index and the preset threshold, and generating an optimization decision of the overall radiation parameters of the multi-frequency antenna unit; S6, executing the optimization decision, dynamically adjusting the overall radiation parameters of the multi-frequency antenna unit, and taking the adjusted wireless signal parameters as feedback data.

2. The intelligent algorithm based method for optimizing the wireless signal coverage of the station platform as claimed in claim 1, wherein, S1 comprises: The multi-frequency antenna unit is configured to simultaneously transmit and receive wireless signals of at least two different mobile communication network modes; And the multi-frequency antenna unit is integrated with an electrically adjustable driving mechanism capable of independently and continuously adjusting the overall radiation parameters.

3. The intelligent algorithm based method for optimizing wireless signal coverage of station platforms as claimed in claim 1 wherein S2 The method comprises: Through the monitoring interface of the multi-frequency antenna unit, the reference signal received power and the signal-to-interference noise ratio of each network mode in the current coverage area are collected in parallel as wireless signal parameters; At the same time, the environmental monitoring data representing passenger flow density and train arrival and departure status are collected through sensors arranged in the station platform environment.

4. The intelligent algorithm based method for optimizing the wireless signal coverage of the station platform as claimed in claim 1, wherein, S3 comprises: The wireless signal parameters and environmental monitoring data of each network mode collected in real time are taken as input to determine the policy gradient update direction of each network mode at the current time with the objective function of improving the network coverage quality of itself; The policy gradient update direction of each network mode is represented as a direction vector in a multi-dimensional parameter space, and the cosine similarity between each pair of direction vectors is calculated; All cosine similarity results are aggregated to generate a scalar value representing the overall policy dispersion degree as a predictive resource allocation conflict degree index.

5. The intelligent algorithm based method for optimizing the wireless signal coverage of the station platform as claimed in claim 1, wherein S4 The method comprises: Based on the predictive resource allocation conflict degree index, the priority direction of each network optimization objective function in the parameter adjustment space is analyzed; The ratio of the overlapping area of the priority direction to the area of each direction exclusive area in the multi-dimensional parameter space is calculated; The ratio is normalized to map a continuous value in the range of 0 to 1, which is used as a policy coordination degree index representing the degree of policy adjustment between networks.

6. The intelligent algorithm based method for optimizing the wireless signal coverage of the station platform as claimed in claim 5, wherein, Based on the predictive resource allocation conflict degree index, the priority direction of each network optimization objective function in the parameter adjustment space is analyzed, which comprises: according to the divergence degree of the policy gradient vector reflected by the predictive resource allocation conflict degree index, the parameter dimension and its adjustment trend most urgently needed to be adjusted in the multi-dimensional parameter space for each network to improve its network coverage quality are reversely inferred to determine the priority direction of each network.

7. The intelligent algorithm based method for optimizing the wireless signal coverage of the station platform according to claim 1, characterized in that S5 The method comprises: When the strategy synergy index is higher than the preset threshold, a multi-objective optimization problem is constructed by using real-time optimization objective functions of each network, a Pareto optimal solution set is searched in a feasible solution space of the overall radiation parameters by a negotiation game algorithm, and an equilibrium solution is selected from the Pareto optimal solution set as a final overall radiation parameter combination to generate an optimization decision; When the strategy synergy index is not higher than the preset threshold, a dominant network is specified according to a preset priority rule, the optimization objective function of the dominant network is set as a main optimization objective, and the overall radiation parameters of other subordinate networks are constrained in an allowed range for single-objective constrained optimization to obtain a solution as a forced synergy scheme and generate an optimization decision.

8. The intelligent algorithm based method for optimizing the wireless signal coverage of the station platform as claimed in claim 7, wherein, The overall radiation parameters include a directional angle, a downtilt angle and an output power.

9. The intelligent algorithm based method for optimizing the wireless signal coverage of the station platform according to claim 1, characterized in that S6 The method comprises the following steps: Converting the overall radiation parameter values determined in the optimization decision into control instruction sequences for internal electric adjustment driving mechanisms of the multi-band antenna unit to drive the multi-band antenna unit to complete adjustment of the directional angle, the downtilt angle and the output power. After the adjustment is completed, the wireless signal parameters of each network fed back by the multi-band antenna unit are collected again, and are encapsulated together with the environmental monitoring data at the corresponding time as new data records as feedback data for generation of the predictive resource allocation conflict degree index in subsequent iterations.

10. A system for implementing the method of any one of claims 1-9, wherein the system is an intelligent algorithm-based system for optimizing wireless signal coverage of a station platform, characterized in that, The method comprises the following steps: A unit installation module is used to arrange a plurality of hanging and holding poles under a station platform, and install a multi-band antenna unit supporting multiple networks on the top of each hanging and holding pole; A data collection module is used to collect wireless signal parameters of each network fed back by the multi-band antenna unit and environmental monitoring data in real time; An index generation module is used to simulate the strategy gradient update direction when each network is an independent optimization intelligent agent based on the wireless signal parameters and the environmental monitoring data, and to calculate the divergence between all strategy gradient update direction vectors to generate a predictive resource allocation conflict degree index; A mutual exclusion evaluation module is used to evaluate the strategy mutual exclusion between real-time optimization objective functions of different networks in the multi-network based on the predictive resource allocation conflict degree index, and to generate a strategy synergy index; A decision generation module is used to selectively call a cooperative game mechanism or a master-slave game mechanism based on a comparison result of the strategy synergy index and a preset threshold to generate an optimization decision of overall radiation parameters of the multi-band antenna unit; A decision execution module is used to execute the optimization decision, dynamically adjust the overall radiation parameters of the multi-band antenna unit, and use the wireless signal parameters generated after the adjustment as feedback data.

Images