Intelligent adjusting method and system based on network load

Abstract

The invention belongs to the technical field of network resource adjustment, and discloses an intelligent adjustment method and system based on a network load. Comprising the following steps: collecting original comprehensive information data of network load equipment; preprocessing the original comprehensive information data to obtain standard comprehensive information data; obtaining a predicted bandwidth utilization rate and a comprehensive fluctuation rate according to the standard comprehensive information data, obtaining a bandwidth allocation priority of the network load equipment according to the standard comprehensive information data, the predicted bandwidth utilization rate and the comprehensive fluctuation rate, carrying out bandwidth allocation according to the bandwidth allocation priority, and carrying out feedback according to a bandwidth allocation result; the allocation of the network bandwidth resources is further optimized, the network bandwidth resources are dynamically adjusted, and the waste of the network bandwidth resources is avoided.

Description

Technical Field

[0001] The present invention relates to the technical field of network resource adjustment, and more specifically, to an intelligent adjustment method and system based on network load. Background Art

[0002] The patent application with the publication number CN116800669A discloses a router-based intelligent load balancing system and method thereof, including a router, a load regulation component, a load regulation block and a feedback component. In this solution, a load balancing system is mounted on the router. When the load regulation component of the active connection on the router is connected to multiple devices through a transmission line, the connected devices are classified and prioritized by the device identification module in the load balancing system. The load balancing control module regulates the connected devices according to the order of priority divided by the device identification module. At the same time, when there is a load regulation conflict, the wireless network control module can control the wireless network to connect to a specific device, thereby allocating data traffic to the best server device, more accurately controlling and managing network traffic and resources, improving network efficiency and reliability, and reducing system failure rate and maintenance costs.

[0003] However, in the process of load balancing, the data is not preprocessed, resulting in outliers in the data, which leads to inaccurate priority division. The impact of changes in historical data of network load devices on the priority division of network load devices is not considered, thereby reducing the accuracy of priority division. After load balancing, no feedback is given to re-divide the priority of network load devices, resulting in subsequent load imbalance.

[0004] In view of this, the present invention proposes an intelligent adjustment method and system based on network load to solve the above problems. Summary of the invention

[0005] In order to overcome the above-mentioned defects of the prior art and to achieve the above-mentioned purpose, the present invention provides the following technical solution: an intelligent adjustment method based on network load, comprising:

[0006] S1: Collect the original comprehensive information data of network load equipment;

[0007] S2: preprocessing the original comprehensive information data to obtain standard comprehensive information data;

[0008] S3: Obtain predicted bandwidth usage and comprehensive volatility according to standard comprehensive information data, obtain bandwidth allocation priority of network load equipment according to standard comprehensive information data, predicted bandwidth usage and comprehensive volatility, allocate bandwidth according to bandwidth allocation priority, and provide feedback according to bandwidth allocation result.

[0009] Furthermore, the original comprehensive information data includes flow data and connection data;

[0010] The traffic data includes bandwidth usage and data transmission rate;

[0011] The connection data includes packet loss rate, memory usage rate and new connection rate.

[0012] Furthermore, the method of preprocessing the original comprehensive information data to obtain standard comprehensive information data includes:

[0013] Step A1: Establish a blank coordinate system, fill n time points into the abscissa of the blank coordinate system in sequence, set the ordinate of the blank coordinate system to the original comprehensive information data, and fill the original comprehensive information data corresponding to the time points into the blank coordinate system to obtain the original comprehensive information data sample points, and then obtain the original comprehensive information data clustering diagram;

[0014] Step A2: obtaining the proximity distance between each original comprehensive information data sample point and the remaining original comprehensive information data sample points according to the coordinates of the original comprehensive information data sample points;

[0015] Step A3: setting a neighbor distance threshold, if the neighbor distance between the original comprehensive information data sample point and the remaining original comprehensive information data sample points is less than or equal to the neighbor distance threshold, then marking the remaining original comprehensive information data sample points as neighboring data sample points of the original comprehensive information data sample point;

[0016] Obtain the number of neighboring data sample points of each original comprehensive information data sample point, and set a number threshold. If the number of neighboring data sample points of the original comprehensive information data sample point is greater than or equal to the number threshold, mark the original comprehensive information data sample point as a normal data sample point;

[0017] Step A4: Obtain original comprehensive information data sample points that are neither marked as adjacent data sample points nor marked as normal data sample points, remove the original comprehensive information data sample points, and use interpolation to fill in the removed original comprehensive information data sample points to obtain standard comprehensive information data.

[0018] Furthermore, the method of obtaining the proximity distance between each original comprehensive information data sample point and the remaining original comprehensive information data sample points according to the coordinates of the original comprehensive information data sample points includes:

[0019] Subtract the horizontal coordinate of each original comprehensive information data sample point from the horizontal coordinates of the remaining original comprehensive information data sample points, subtract the vertical coordinate of each original comprehensive information data sample point from the vertical coordinates of the remaining original comprehensive information data sample points, take the absolute value of the two subtraction results and add them together to obtain the broken line distance between each original comprehensive information data sample point and the remaining original comprehensive information data sample points;

[0020] Obtain the straight-line distance between each original comprehensive information data sample point and the remaining original comprehensive information data sample points, take the average of the broken line distance and the straight-line distance, and obtain the proximity distance between each original comprehensive information data sample point and the remaining original comprehensive information data sample points.

[0021] Furthermore, the method for obtaining predicted bandwidth usage rate according to standard comprehensive information data includes:

[0022] Build and train a bandwidth usage prediction model, including:

[0023] Based on the LSTM basic framework, the input layer, LSTM layer, fully connected layer and output layer are set. The input of the input layer is set to standard comprehensive information data, the activation function of the fully connected layer is ReLU, and the output of the output layer is the predicted bandwidth usage rate.

[0024] Using the Adam algorithm, the loss function is: y(s_p) represents the predicted bandwidth usage of the (s_p+1)th sample, y'(s_p) represents the actual bandwidth usage of the (s_p+1)th sample, SL is the number of samples, ε is the weight coefficient, and s_p is the index of the number of samples;

[0025] The training and use of the bandwidth usage prediction model include:

[0026] Step B1: Collect the sample set of group p_o and divide it into training set and validation set in the ratio of 8:2. Input the training set into batches and input them into the bandwidth usage prediction model in turn for forward propagation.

[0027] Step B2: Obtain the predicted bandwidth usage output by the bandwidth usage prediction model, calculate the loss value using the loss function, calculate each parameter in the model using the back propagation algorithm, and update the parameters using the gradient descent algorithm;

[0028] Step B3: Repeat steps B1 and B2 until the loss function value of the bandwidth usage prediction model no longer changes, import the validation set for validation, and when the validation is successful, obtain the trained bandwidth usage prediction model; when the validation is unsuccessful, repeat step B3;

[0029] Step B4: Input the standard comprehensive information data at the nth time point into the trained bandwidth utilization rate prediction model to obtain the predicted bandwidth utilization rate at the next time point after the nth time point.

[0030] Further, the method for obtaining the comprehensive volatility based on the standard comprehensive information data includes:

[0031] Step C1: Subtract the bandwidth utilization rate at the previous time point from the bandwidth utilization rate at the next time point, obtain the absolute value of all subtraction results, and calculate the average value of the absolute values of all subtraction results to obtain the bandwidth volatility;

[0032] Step C2: Repeat Step C1 to process the data transmission rate, packet loss rate, memory utilization rate, and new connection rate respectively, and obtain the transmission volatility, packet loss volatility, memory volatility, and connection volatility respectively;

[0033] Step C3: Take the average value of the bandwidth volatility, transmission volatility, packet loss volatility, memory volatility, and connection volatility to obtain the comprehensive volatility.

[0034] Further, the method for obtaining the bandwidth allocation priority of the network load device based on the standard comprehensive information data, predicted bandwidth utilization rate, and comprehensive volatility includes:

[0035] The bandwidth allocation priority includes the first bandwidth priority, the second bandwidth priority, and the third bandwidth priority;

[0036] Set the bandwidth utilization rate threshold range (YDK0, YDK1);

[0037] Pre-allocate the bandwidth allocation priority of the network load device according to the comparison result between the predicted bandwidth utilization rate and the bandwidth utilization rate threshold range;

[0038] When YDK ≤ YDK0, pre-allocate the bandwidth allocation priority of the network load device as the third bandwidth priority;

[0039] When YDK0 < YDK < YDK1, pre-allocate the bandwidth allocation priority of the network load device as the second bandwidth priority;

[0040] When YDK ≥ YDK1, pre-allocate the bandwidth allocation priority of the network load device as the first bandwidth priority;

[0041] Wherein, YDK is the predicted bandwidth utilization rate at the next time point after the nth time point;

[0042] Multiply the comprehensive volatility by the standard comprehensive information data at the nth time point and the predicted bandwidth utilization rate at the next time point after the nth time point respectively, and add the multiplied results to obtain the priority adjustment coefficient YTZ;

[0043] Set the threshold range (YTZ0, YTZ1) of the priority adjustment coefficient;

[0044] When YTZ ≤ YTZ0, if the bandwidth allocation priority of the network load device is pre-allocated as the first bandwidth priority, then set the bandwidth allocation priority of the network load device as the second bandwidth priority; if the bandwidth allocation priority of the network load device is pre-allocated as the second bandwidth priority, then set the bandwidth allocation priority of the network load device as the third bandwidth priority; if the bandwidth allocation priority of the network load device is pre-allocated as the third bandwidth priority, then keep the bandwidth allocation priority of the network load device unchanged;

[0045] When YTZ0 < YTZ < YTZ1, keep the bandwidth allocation priority of the network load device unchanged;

[0046] When YTZ ≥ YTZ1, if the bandwidth allocation priority of the network load device is pre-allocated as the first bandwidth priority or the second bandwidth priority, then keep the bandwidth allocation priority of the network load device unchanged; if the bandwidth allocation priority of the network load device is pre-allocated as the third bandwidth priority, then set the bandwidth allocation priority of the network load device as the second bandwidth priority.

[0047] Further, the method for bandwidth allocation according to the bandwidth allocation priority includes:

[0048] If the bandwidth allocation priority of the network load device is the first bandwidth priority, then the size of the bandwidth allocated to this network load device is s_dk1; if the bandwidth allocation priority of the network load device is the second bandwidth priority, then the size of the bandwidth allocated to this network load device is s_dk2; if the bandwidth allocation priority of the network load device is the third bandwidth priority, then the size of the bandwidth allocated to this network load device is s_dk3;

[0049] where s_dk1 > s_dk2 > s_dk3.

[0050] Further, the method for feedback according to the bandwidth allocation result includes:

[0051] Obtain the bandwidth utilization rate of the network load device after bandwidth allocation and record it as the feedback utilization rate;

[0052] When the feedback utilization rate is greater than or equal to the bandwidth utilization rate at the nth time point, keep the bandwidth allocation priority of the network load device unchanged;

[0053] When the feedback utilization rate is less than the bandwidth utilization rate at the nth time point, obtain the comprehensive volatility of the network load device after bandwidth allocation, and multiply the comprehensive volatility of the network load device after bandwidth allocation by the standard comprehensive information data at the (n + 1)th time point to obtain the feedback adjustment coefficient YFK;

[0054] Set the feedback adjustment coefficient threshold range (YFK0, YFK1);

[0055] When YFK ≤ YFK0, if the bandwidth allocation priority of the network load device is the first bandwidth priority, then set the bandwidth allocation priority of the network load device to the second bandwidth priority; if the bandwidth allocation priority of the network load device is the second bandwidth priority, then set the bandwidth allocation priority of the network load device to the third bandwidth priority; if the bandwidth allocation priority of the network load device is the third bandwidth priority, then keep the bandwidth allocation priority of the network load device unchanged;

[0056] When YFK0 < YFK < YFK1, keep the bandwidth allocation priority of the network load device unchanged;

[0057] When YFK ≥ YFK1, if the bandwidth allocation priority of the network load device is the first bandwidth priority or the second bandwidth priority, then keep the bandwidth allocation priority of the network load device unchanged; if the bandwidth allocation priority of the network load device is the third bandwidth priority, then set the bandwidth allocation priority of the network load device to the second bandwidth priority.

[0058] The intelligent regulation system based on network load includes:

[0059] A data acquisition module, which is used to acquire the original comprehensive information data of the network load device;

[0060] A data preprocessing module, which is used to preprocess the original comprehensive information data to obtain the standard comprehensive information data;

[0061] An intelligent regulation module, which is used to obtain the bandwidth allocation priority of the network load device according to the standard comprehensive information data, perform bandwidth allocation according to the bandwidth allocation priority, and give feedback according to the bandwidth allocation result.

[0062] The technical effects and advantages of the intelligent regulation method and system based on network load of the present invention:

[0063] In the present invention, by preprocessing the original comprehensive information data to obtain standard comprehensive information data, the abnormal values ​​in the original comprehensive information data are eliminated and the normal values ​​are filled, so that the collected original comprehensive information data is more accurate, the predicted bandwidth utilization rate and the comprehensive volatility are obtained according to the standard comprehensive information data, the bandwidth allocation priority of the network load device is pre-allocated according to the predicted bandwidth utilization rate, and the accuracy of the subsequent bandwidth allocation priority division is further improved, the bandwidth allocation priority of the network load device is obtained according to the standard comprehensive information data, the predicted bandwidth utilization rate and the comprehensive volatility, and the influence of the fluctuation of the standard comprehensive information data on the bandwidth allocation priority division is considered, so that the bandwidth allocation priority division is more accurate, and the bandwidth is allocated according to the bandwidth allocation priority, and feedback is given according to the bandwidth allocation result, so that the subsequent bandwidth allocation is more balanced. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 A schematic diagram of an intelligent adjustment method based on network load according to the present invention;

[0065] Figure 2 It is a schematic diagram of the intelligent regulation system based on network load of the present invention. DETAILED DESCRIPTION

[0066] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0067] Example 1

[0068] See also Figure 1 As shown, the intelligent adjustment method based on network load described in this embodiment includes:

[0069] S1: Collect the original comprehensive information data of network load equipment;

[0070] S2: preprocessing the original comprehensive information data to obtain standard comprehensive information data;

[0071] S3: Obtain predicted bandwidth usage and comprehensive volatility according to standard comprehensive information data, obtain bandwidth allocation priority of network load equipment according to standard comprehensive information data, predicted bandwidth usage and comprehensive volatility, allocate bandwidth according to bandwidth allocation priority, and provide feedback according to bandwidth allocation result;

[0072] The process of collecting the original comprehensive information data of network load devices includes:

[0073] The original comprehensive information data includes flow data and connection data;

[0074] The traffic data includes bandwidth usage and data transmission rate, and the connection data includes packet loss rate, memory usage rate and new connection rate;

[0075] The bandwidth utilization rate is the ratio of the actual bandwidth utilization of the network load device to the total available bandwidth of the network load device multiplied by 100%;

[0076] The data transmission rate is the sum of the number of bytes and the number of data packets transmitted per second by the network load device;

[0077] The packet loss rate is the ratio of the number of data packets lost by the network load device during data transmission to the total number of data packets sent multiplied by 100%;

[0078] The memory usage rate is the ratio of the used memory amount of the network load device to the total available memory amount multiplied by 100%;

[0079] The new connection rate is the number of new connections established by the network load device per second;

[0080] The configuration data acquisition terminal generates a corresponding configuration channel, and the network load device is linked through the configuration channel. The data acquisition terminal obtains the original comprehensive information data of the network load device through the configuration channel. The data acquisition terminal is set with a fixed acquisition cycle.

[0081] The process of preprocessing the original comprehensive information data to obtain standard comprehensive information data includes:

[0082] Step A1: Establish a blank coordinate system, fill n time points into the abscissa of the blank coordinate system in sequence, set the ordinate of the blank coordinate system to the original comprehensive information data, and fill the original comprehensive information data corresponding to the time points into the blank coordinate system to obtain the original comprehensive information data sample points, and then obtain the original comprehensive information data clustering diagram;

[0083] Step A2: subtract the abscissa of each original comprehensive information data sample point from the abscissa of the remaining original comprehensive information data sample points, subtract the ordinate of each original comprehensive information data sample point from the ordinate of the remaining original comprehensive information data sample points, take the absolute value of the two subtraction results and add them together to obtain the broken line distance between each original comprehensive information data sample point and the remaining original comprehensive information data sample points, obtain the straight line distance between each original comprehensive information data sample point and the remaining original comprehensive information data sample points, take the average of the broken line distance and the straight line distance, and obtain the neighboring distance between each original comprehensive information data sample point and the remaining original comprehensive information data sample points;

[0084] For example, if the coordinates of two original comprehensive information data sample points are (x1, y1) and (x2, y2), the polyline distance between the two original comprehensive information data sample points is: ZX = |x1-x2| + |y1-y2|; ZX is the polyline distance between the two original comprehensive information data sample points, and the straight-line distance between the two original comprehensive information data sample points is: HX is the straight-line distance between two original comprehensive information data sample points, and the proximity distance between two original comprehensive information data sample points is: LJ is the proximity distance between two original comprehensive information data sample points;

[0085] Step A3: setting a neighboring distance threshold, if the neighboring distance between the original comprehensive information data sample point and the remaining original comprehensive information data sample points is less than or equal to the neighboring distance threshold, then marking the remaining original comprehensive information data sample points as neighboring data sample points of the original comprehensive information data sample point, obtaining the number of neighboring data sample points of each original comprehensive information data sample point, and setting a number threshold, if the number of neighboring data sample points of the original comprehensive information data sample point is greater than or equal to the number threshold, then marking the original comprehensive information data sample point as a normal data sample point;

[0086] Step A4: Obtain original comprehensive information data sample points that are neither marked as adjacent data sample points nor marked as normal data sample points, remove the original comprehensive information data sample points, and use interpolation to fill in the removed original comprehensive information data sample points to obtain standard comprehensive information data.

[0087] The process of obtaining predicted bandwidth usage based on standard comprehensive information data includes:

[0088] Build and train a bandwidth usage prediction model, including:

[0089] Based on the LSTM basic framework, the input layer, LSTM layer, fully connected layer and output layer are set. The input of the input layer is set to standard comprehensive information data, the activation function of the fully connected layer is ReLU, and the output of the output layer is the predicted bandwidth usage rate.

[0090] The Adam algorithm is used, the learning rate is set to 0.001, and the loss function is: y(s_p) represents the predicted bandwidth usage of the (s_p+1)th sample, y'(s_p) represents the actual bandwidth usage of the (s_p+1)th sample, SL is the number of samples, ε is the weight coefficient, and s_p is the index of the number of samples;

[0091] The training and use of the bandwidth usage prediction model include:

[0092] Step B1: Collect the sample set of group p_o and divide it into training set and validation set in the ratio of 8:2. Input the training set into batches and input them into the bandwidth usage prediction model in turn for forward propagation.

[0093] Step B2: Obtain the predicted bandwidth usage output by the bandwidth usage prediction model, calculate the loss value using the loss function, calculate each parameter in the model using the back propagation algorithm, and update the parameters using the gradient descent algorithm;

[0094] Step B3: Repeat steps B1 and B2 until the loss function value of the bandwidth usage prediction model no longer changes, import the validation set for validation, and when the validation is successful, obtain the trained bandwidth usage prediction model; when the validation is unsuccessful, repeat step B3;

[0095] Step B4: inputting the standard comprehensive information data at the current time point into the trained bandwidth usage prediction model to obtain the predicted bandwidth usage YDK at the next time point of the current time point;

[0096] The sample set settings include:

[0097] Obtain standard comprehensive information data at (p_o+1) time points, take the standard comprehensive information data at one time point and the standard comprehensive information data at the next time point as a sample, and collect all samples at p_o time points to form a sample set.

[0098] The process of obtaining the composite volatility based on the standard composite information data includes:

[0099] Subtract the bandwidth usage rate at the previous time point from the bandwidth usage rate at the next time point, obtain the absolute values ​​of all subtraction results, and calculate the average value of the absolute values ​​of all subtraction results to obtain the bandwidth fluctuation rate DB;

[0100] in, i is the index of the time point, n is the number of time points, db i is the bandwidth usage at the i-th time point, db i-1 is the bandwidth usage rate at the previous time point i-1 of the i-th time point;

[0101] Subtract the data transmission rate at the previous time point from the data transmission rate at the next time point, obtain the absolute value of all subtraction results, and calculate the average value of the absolute value of all subtraction results to obtain the transmission fluctuation rate CB;

[0102] in, cb i is the data transmission rate at the i-th time point, cb i-1is the data transmission rate at the previous time point i-1 of the i-th time point;

[0103] Subtract the packet loss rate at the previous time point from the packet loss rate at the next time point, obtain the absolute value of all subtraction results, and calculate the average of the absolute values ​​of all subtraction results to obtain the packet loss volatility IB;

[0104] in, ib i is the packet loss rate at the i-th time point, ib i-1 is the packet loss rate at the previous time point i-1 of the i-th time point;

[0105] Subtract the memory usage rate at the previous time point from the memory usage rate at the next time point, obtain the absolute value of all subtraction results, and calculate the average of the absolute values ​​of all subtraction results to obtain the memory volatility NB;

[0106] in, nb i is the memory usage at the i-th time point, nb i-1 is the memory usage at the previous time point i-1 of the i-th time point;

[0107] Subtract the new connection rate at the previous time point from the new connection rate at the next time point, obtain the absolute value of all subtraction results, and calculate the average of the absolute values ​​of all subtraction results to obtain the connection volatility LB;

[0108] in, lb i is the new connection rate at the i-th time point, lb i-1 is the new connection rate at the previous time point i-1 of the i-th time point;

[0109] Take the average of bandwidth volatility, transmission volatility, packet loss volatility, memory volatility and connection volatility to obtain the comprehensive volatility ZB;

[0110] in,

[0111] The process of obtaining bandwidth allocation priorities for network load devices based on standard comprehensive information data, predicted bandwidth usage, and comprehensive fluctuation rate includes:

[0112] The bandwidth allocation priority includes a first bandwidth priority, a second bandwidth priority and a third bandwidth priority;

[0113] It should be explained that the bandwidth allocated by the first bandwidth priority is higher than the bandwidth allocated by the second bandwidth priority, which is higher than the bandwidth allocated by the third bandwidth priority;

[0114] Set the bandwidth utilization threshold range (YDK0, YDK1);

[0115] Pre - allocate the bandwidth allocation priority of the network load device according to the comparison result between the predicted bandwidth utilization and the bandwidth utilization threshold range;

[0116] When YDK ≤ YDK0, pre - allocate the bandwidth allocation priority of the network load device as the third bandwidth priority;

[0117] When YDK0 < YDK < YDK1, pre - allocate the bandwidth allocation priority of the network load device as the second bandwidth priority;

[0118] When YDK ≥ YDK1, pre - allocate the bandwidth allocation priority of the network load device as the first bandwidth priority;

[0119] Obtain the priority adjustment coefficient YTZ according to the comprehensive volatility and the standard comprehensive information data;

[0120] Where, YTZ = ZB×(db n +cb n +ib n +nb n +lb n +YDK); db n is the bandwidth utilization at the n - th time point, cb n is the data transmission rate at the n - th time point, ib n is the packet loss rate at the n - th time point, nb n is the memory utilization at the n - th time point, lb n is the new connection rate at the n - th time point;

[0121] Set the priority adjustment coefficient threshold range (YTZ0, YTZ1);

[0122] When YTZ ≤ YTZ0, if the bandwidth allocation priority of the network load device is pre - allocated as the first bandwidth priority, then set the bandwidth allocation priority of the network load device as the second bandwidth priority; if the bandwidth allocation priority of the network load device is pre - allocated as the second bandwidth priority, then set the bandwidth allocation priority of the network load device as the third bandwidth priority; if the bandwidth allocation priority of the network load device is pre - allocated as the third bandwidth priority, then keep the bandwidth allocation priority of the network load device unchanged;

[0123] When YTZ0 < YTZ < YTZ1, if the bandwidth allocation priority of the network load device is pre-allocated as the first bandwidth priority, the bandwidth allocation priority of the network load device remains unchanged; if the bandwidth allocation priority of the network load device is pre-allocated as the second bandwidth priority, the bandwidth allocation priority of the network load device remains unchanged; if the bandwidth allocation priority of the network load device is pre-allocated as the third bandwidth priority, the bandwidth allocation priority of the network load device remains unchanged;

[0124] When YTZ ≥ YTZ1, if the bandwidth allocation priority of the network load device is pre-allocated as the first bandwidth priority, the bandwidth allocation priority of the network load device remains unchanged; if the bandwidth allocation priority of the network load device is pre-allocated as the second bandwidth priority, the bandwidth allocation priority of the network load device remains unchanged; if the bandwidth allocation priority of the network load device is pre-allocated as the third bandwidth priority, the bandwidth allocation priority of the network load device is set to the second bandwidth priority;

[0125] The process of bandwidth allocation according to the bandwidth allocation priority includes:

[0126] If the bandwidth allocation priority of the network load device is the first bandwidth priority, the size of the bandwidth allocated to this network load device is s_dk1; if the bandwidth allocation priority of the network load device is the second bandwidth priority, the size of the bandwidth allocated to this network load device is s_dk2; if the bandwidth allocation priority of the network load device is the third bandwidth priority, the size of the bandwidth allocated to this network load device is s_dk3;

[0127] It should be noted that s_dk1 > s_dk2 > s_dk3;

[0128] The process of feedback according to the bandwidth allocation result includes:

[0129] Obtain the bandwidth utilization rate of the network load device after bandwidth allocation and record it as the feedback utilization rate;

[0130] When the feedback utilization rate is greater than or equal to the bandwidth utilization rate at the nth time point, the bandwidth allocation priority of the network load device remains unchanged;

[0131] When the feedback utilization rate is less than the bandwidth utilization rate at the nth time point, obtain the comprehensive volatility of the network load device after bandwidth allocation, and obtain the feedback adjustment coefficient YFK according to the comprehensive volatility of the network load device after bandwidth allocation;

[0132] Among them, YFK = DZB × (db n+1 +cb n+1 +ib n+1 +nb n+1 +lb n+1); DZB is the comprehensive volatility of the network load device after bandwidth allocation, in db n+1 is the bandwidth utilization rate at the (n + 1)-th time point, in cb n+1 is the data transmission rate at the (n + 1)-th time point, in ib n+1 is the packet loss rate at the (n + 1)-th time point, in nb n+1 is the memory utilization rate at the (n + 1)-th time point, in lb n+1 is the new connection rate at the (n + 1)-th time point;

[0133] Set the feedback adjustment coefficient threshold range (YFK0, YFK1);

[0134] When YFK ≤ YFK0, if the bandwidth allocation priority of the network load device is the first bandwidth priority, then set the bandwidth allocation priority of the network load device to the second bandwidth priority. If the bandwidth allocation priority of the network load device is the second bandwidth priority, then set the bandwidth allocation priority of the network load device to the third bandwidth priority. If the bandwidth allocation priority of the network load device is the third bandwidth priority, then keep the bandwidth allocation priority of the network load device unchanged;

[0135] When YFK0 < YFK < YFK1, keep the bandwidth allocation priority of the network load device unchanged;

[0136] When YFK ≥ YFK1, if the bandwidth allocation priority of the network load device is the first bandwidth priority or the second bandwidth priority, then keep the bandwidth allocation priority of the network load device unchanged. If the bandwidth allocation priority of the network load device is the third bandwidth priority, then set the bandwidth allocation priority of the network load device to the second bandwidth priority.

[0137] In this embodiment, by preprocessing the original comprehensive information data, standard comprehensive information data is obtained. Outliers in the original comprehensive information data are removed, and normal values are filled, making the collected original comprehensive information data more accurate. The predicted bandwidth utilization rate and comprehensive volatility are obtained based on the standard comprehensive information data. The bandwidth allocation priority of the network load device is pre-allocated according to the predicted bandwidth utilization rate, further improving the accuracy of subsequent bandwidth allocation priority division. The bandwidth allocation priority of the network load device is obtained based on the standard comprehensive information data, predicted bandwidth utilization rate, and comprehensive volatility, considering the impact of fluctuations in the standard comprehensive information data on the bandwidth allocation priority division, making the division of bandwidth allocation priority more accurate. Bandwidth allocation is performed according to the bandwidth allocation priority, and feedback is carried out based on the bandwidth allocation result, making subsequent bandwidth allocation more balanced.

[0138] Embodiment 2

[0139] Please refer to Figure 2As shown, the part not described in detail in this embodiment is described in Example 1, and an intelligent adjustment system based on network load is provided, including:

[0140] Data acquisition module, used to collect original comprehensive information data of network load equipment;

[0141] A data preprocessing module is used to preprocess the original comprehensive information data to obtain standard comprehensive information data;

[0142] The intelligent adjustment module is used to obtain the bandwidth allocation priority of the network load device according to the standard comprehensive information data, allocate bandwidth according to the bandwidth allocation priority, and provide feedback according to the bandwidth allocation result.

[0143] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed in the present invention can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0144] In the several embodiments provided by the present invention, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of the units is only one, and there may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0145] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

[0146] Finally: 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 in the protection scope of the present invention.

Claims

1. An intelligent adjustment method based on network load, characterized in that: The intelligent adjustment method based on network load includes: S1: Collect the original comprehensive information data of network load equipment; S2: preprocessing the original comprehensive information data to obtain standard comprehensive information data; S3: Obtain predicted bandwidth usage and comprehensive volatility according to standard comprehensive information data, obtain bandwidth allocation priority of network load equipment according to standard comprehensive information data, predicted bandwidth usage and comprehensive volatility, allocate bandwidth according to bandwidth allocation priority, and provide feedback according to bandwidth allocation result.

2. The intelligent adjustment method based on network load according to claim 1 is characterized in that: The original comprehensive information data includes flow data and connection data; The traffic data includes bandwidth usage and data transmission rate; The connection data includes packet loss rate, memory usage rate and new connection rate.

3. The intelligent adjustment method based on network load according to claim 2 is characterized in that: The method of preprocessing the original comprehensive information data to obtain standard comprehensive information data includes: Step A1: Establish a blank coordinate system, fill n time points into the abscissa of the blank coordinate system in sequence, set the ordinate of the blank coordinate system to the original comprehensive information data, and fill the original comprehensive information data corresponding to the time points into the blank coordinate system to obtain the original comprehensive information data sample points, and then obtain the original comprehensive information data clustering diagram; Step A2: obtaining the proximity distance between each original comprehensive information data sample point and the remaining original comprehensive information data sample points according to the coordinates of the original comprehensive information data sample points; Step A3: setting a neighbor distance threshold, if the neighbor distance between the original comprehensive information data sample point and the remaining original comprehensive information data sample points is less than or equal to the neighbor distance threshold, then marking the remaining original comprehensive information data sample points as neighboring data sample points of the original comprehensive information data sample point; Obtain the number of neighboring data sample points of each original comprehensive information data sample point, and set a number threshold. If the number of neighboring data sample points of the original comprehensive information data sample point is greater than or equal to the number threshold, mark the original comprehensive information data sample point as a normal data sample point; Step A4: Obtain original comprehensive information data sample points that are neither marked as adjacent data sample points nor marked as normal data sample points, remove the original comprehensive information data sample points, and use interpolation to fill in the removed original comprehensive information data sample points to obtain standard comprehensive information data.

4. The intelligent adjustment method based on network load according to claim 3 is characterized in that: The method for obtaining the proximity distance between each original comprehensive information data sample point and the remaining original comprehensive information data sample points according to the coordinates of the original comprehensive information data sample point comprises: Subtract the horizontal coordinate of each original comprehensive information data sample point from the horizontal coordinates of the remaining original comprehensive information data sample points, subtract the vertical coordinate of each original comprehensive information data sample point from the vertical coordinates of the remaining original comprehensive information data sample points, take the absolute value of the two subtraction results and add them together to obtain the broken line distance between each original comprehensive information data sample point and the remaining original comprehensive information data sample points; Obtain the straight-line distance between each original comprehensive information data sample point and the remaining original comprehensive information data sample points, take the average of the broken line distance and the straight-line distance, and obtain the proximity distance between each original comprehensive information data sample point and the remaining original comprehensive information data sample points.

5. The intelligent adjustment method based on network load according to claim 4 is characterized in that: The method for obtaining the predicted bandwidth utilization rate based on standard comprehensive information data includes: Constructing and training a bandwidth utilization rate prediction model, specifically including: Based on the basic LSTM framework, setting an input layer, an LSTM layer, a fully connected layer, and an output layer, setting the input of the input layer as the standard comprehensive information data, the activation function of the fully connected layer as ReLU, and the output of the output layer as the predicted bandwidth utilization rate; Using the Adam algorithm, the loss function is: y(s_p) represents the predicted bandwidth usage of the (s_p+1)th sample, y'(s_p) represents the actual bandwidth usage of the (s_p+1)th sample, SL is the number of samples, ε is the weight coefficient, and s_p is the index of the number of samples; The training and use of the bandwidth utilization rate prediction model include: Step B1: Collect p_o groups of sample sets, divide them into a training set and a validation set according to a ratio of 8:2, and sequentially input the training set in batches into the bandwidth utilization rate prediction model for forward propagation; Step B2: Obtain the predicted bandwidth utilization rate output by the bandwidth utilization rate prediction model, calculate the loss value using a loss function, calculate each parameter in the model through the backpropagation algorithm, and update the parameters using the gradient descent algorithm; Step B3: Repeat steps B1 and B2 until the loss function value of the bandwidth utilization rate prediction model no longer changes, then import the validation set for validation. When the validation is successful, obtain the trained bandwidth utilization rate prediction model. When the validation is unsuccessful, repeat step B3; Step B4: Input the standard comprehensive information data at the nth time point into the trained bandwidth utilization rate prediction model to obtain the predicted bandwidth utilization rate at the next time point after the nth time point.

6. The intelligent adjustment method based on network load according to claim 5 is characterized in that: The method for obtaining the comprehensive volatility based on the standard comprehensive information data includes: Step C1: Subtract the bandwidth utilization rate at the previous time point from the bandwidth utilization rate at the next time point, obtain the absolute value of all subtraction results, and calculate the average value of the absolute values of all subtraction results to obtain the bandwidth volatility; Step C2: Repeat step C1 to process the data transmission rate, packet loss rate, memory utilization rate, and new connection rate respectively, and obtain the transmission volatility, packet loss volatility, memory volatility, and connection volatility respectively; Step C3: Take the average value of the bandwidth volatility, transmission volatility, packet loss volatility, memory volatility, and connection volatility to obtain the comprehensive volatility.

7. The intelligent adjustment method based on network load according to claim 6, characterized in that: The method for obtaining the bandwidth allocation priority of network load devices based on the standard comprehensive information data, predicted bandwidth utilization rate, and comprehensive volatility includes: The bandwidth allocation priority includes the first bandwidth priority, the second bandwidth priority, and the third bandwidth priority; Set the bandwidth utilization rate threshold range (YDK0, YDK1); Pre-allocate the bandwidth allocation priority of the network load device according to the comparison result between the predicted bandwidth utilization rate and the bandwidth utilization rate threshold range; When YDK ≤ YDK0, pre-allocate the bandwidth allocation priority of the network load device as the third bandwidth priority; When YDK0 < YDK < YDK1, pre-allocate the bandwidth allocation priority of the network load device as the second bandwidth priority; When YDK ≥ YDK1, pre-allocate the bandwidth allocation priority of the network load device as the first bandwidth priority; where YDK is the predicted bandwidth utilization rate at the next time point after the nth time point; Multiply the comprehensive volatility by the standard comprehensive information data at the nth time point and the predicted bandwidth utilization rate at the next time point after the nth time point respectively, and add the multiplied results to obtain the priority adjustment coefficient YTZ; Set the threshold range of the priority adjustment coefficient (YTZ0, YTZ1); When YTZ ≤ YTZ0, if the bandwidth allocation priority of the network load device is pre-allocated as the first bandwidth priority, then set the bandwidth allocation priority of the network load device as the second bandwidth priority; if the bandwidth allocation priority of the network load device is pre-allocated as the second bandwidth priority, then set the bandwidth allocation priority of the network load device as the third bandwidth priority; if the bandwidth allocation priority of the network load device is pre-allocated as the third bandwidth priority, then keep the bandwidth allocation priority of the network load device unchanged; When YTZ0 < YTZ < YTZ1, keep the bandwidth allocation priority of the network load device unchanged; When YTZ ≥ YTZ1, if the bandwidth allocation priority of the network load device is pre-allocated as the first bandwidth priority or the second bandwidth priority, then keep the bandwidth allocation priority of the network load device unchanged; if the bandwidth allocation priority of the network load device is pre-allocated as the third bandwidth priority, then set the bandwidth allocation priority of the network load device as the second bandwidth priority.

8. The intelligent adjustment method based on network load according to claim 7 is characterized in that: The method for bandwidth allocation according to the bandwidth allocation priority includes: If the bandwidth allocation priority of the network load device is the first bandwidth priority, then the size of the bandwidth allocated to this network load device is s_dk1; if the bandwidth allocation priority of the network load device is the second bandwidth priority, then the size of the bandwidth allocated to this network load device is s_dk2; if the bandwidth allocation priority of the network load device is the third bandwidth priority, then the size of the bandwidth allocated to this network load device is s_dk3; where s_dk1 > s_dk2 > s_dk3.

9. The intelligent adjustment method based on network load according to claim 8, characterized in that: The method for feedback according to the bandwidth allocation result includes: Obtain the bandwidth utilization rate of the network load device after bandwidth allocation and record it as the feedback utilization rate; When the feedback utilization rate is greater than or equal to the bandwidth utilization rate at the nth time point, keep the bandwidth allocation priority of the network load device unchanged; When the feedback utilization rate is less than the bandwidth utilization rate at the nth time point, obtain the comprehensive volatility of the network load device after bandwidth allocation, and multiply the comprehensive volatility of the network load device after bandwidth allocation by the standard comprehensive information data at the (n + 1)th time point to obtain the feedback adjustment coefficient YFK; Set the threshold range of the feedback adjustment coefficient (YFK0, YFK1); When YFK ≤ YFK0, if the bandwidth allocation priority of the network load device is the first bandwidth priority, then set the bandwidth allocation priority of the network load device as the second bandwidth priority; if the bandwidth allocation priority of the network load device is the second bandwidth priority, then set the bandwidth allocation priority of the network load device as the third bandwidth priority; if the bandwidth allocation priority of the network load device is the third bandwidth priority, then keep the bandwidth allocation priority of the network load device unchanged; When YFK0 < YFK < YFK1, keep the bandwidth allocation priority of the network load device unchanged; When YFK ≥ YFK1, if the bandwidth allocation priority of the network load device is the first bandwidth priority or the second bandwidth priority, keep the bandwidth allocation priority of the network load device unchanged. If the bandwidth allocation priority of the network load device is the third bandwidth priority, set the bandwidth allocation priority of the network load device to the second bandwidth priority.

10. An intelligent regulation system based on network load, which is used to implement the intelligent regulation method based on network load according to any one of claims 1 to 9, characterized in that: Including: A data acquisition module for acquiring the original comprehensive information data of the network load device; A data preprocessing module for preprocessing the original comprehensive information data to obtain standard comprehensive information data; An intelligent adjustment module for obtaining the bandwidth allocation priority of the network load device according to the standard comprehensive information data, performing bandwidth allocation according to the bandwidth allocation priority, and giving feedback according to the bandwidth allocation result.

Images