Business data processing methods, devices, equipment, storage media, and program products

Abstract

This application relates to a business data processing method, apparatus, device, storage medium, and program product. The method includes: receiving a business processing instruction; determining a target nonlinear function and the values ​​of its independent variables based on the instruction, the instruction indicating processing of a communication signal; determining a target independent variable interval from multiple candidate independent variable intervals corresponding to the target nonlinear function, wherein the multiple candidate intervals are obtained by non-uniformly dividing the range of independent variable values ​​of the target nonlinear function, and the maximum fitting error corresponding to each candidate interval is equal; determining a target linear fitting coefficient corresponding to the target independent variable interval; and processing the communication signal according to the target linear fitting coefficient in response to the instruction. This method can reduce computational complexity and improve operational efficiency.

Description

Technical Field

[0001] This application relates to the field of data processing technology, and in particular to a business data processing method, apparatus, device, storage medium, and program product. Background Technology

[0002] During operation, the terminal relies on chips within it to perform various calculations to achieve related communication functions. For example, the chips in the terminal may include demodulation modules, decoding modules, radio frequency modules, or channel estimation modules. Each module may involve different nonlinear functions when processing communication signals. Due to hardware implementation limitations, the values ​​of nonlinear functions cannot be directly calculated and must be obtained through fitting.

[0003] In related technologies, for a target nonlinear function to be fitted, the range of values ​​of the independent variable of the target nonlinear function is evenly divided into multiple sub-intervals, and the fitting coefficients corresponding to each sub-interval are determined based on the fitting error of each sub-interval. Thus, during the operation, the chip determines the value of the target nonlinear function based on the pre-determined fitting coefficients.

[0004] However, in related technologies, dividing the range of independent variables into equal parts results in a large number of sub-intervals, which in turn leads to a large number of fitting coefficients. Therefore, the computational complexity of obtaining fitting coefficients during chip operation is high, the amount of computation is large, the running time is long, and the fitting efficiency is affected. Summary of the Invention

[0005] Therefore, it is necessary to provide a business data processing method, apparatus, device, storage medium, and program product that can reduce computational complexity and improve computational efficiency in response to the above-mentioned technical problems.

[0006] Firstly, this application provides a business data processing method. The method includes:

[0007] Receive a service processing instruction, determine the target nonlinear function and the value of the independent variable of the target nonlinear function according to the service processing instruction, and the service processing instruction is used to instruct the processing of the communication signal;

[0008] From the multiple candidate independent variable intervals corresponding to the target nonlinear function, the target independent variable interval in which the value of the independent variable is located is determined. The multiple candidate independent variable intervals are obtained by non-uniformly dividing the range of independent variable values ​​of the target nonlinear function, and the maximum fitting error corresponding to each candidate independent variable interval is equal.

[0009] Determine the target linear fitting coefficients corresponding to the target independent variable interval, and process the communication signal in response to the service processing instruction based on the target linear fitting coefficients.

[0010] Secondly, this application also provides a business data processing apparatus. The apparatus includes:

[0011] The first determining module is used to receive a service processing instruction, determine the target nonlinear function and the value of the independent variable of the target nonlinear function according to the service processing instruction, and the service processing instruction is used to instruct the processing of the communication signal;

[0012] The second determining module is used to determine the target independent variable interval where the value of the independent variable is located from multiple candidate independent variable intervals corresponding to the target nonlinear function. The multiple candidate independent variable intervals are obtained by non-uniformly dividing the range of independent variable values ​​of the target nonlinear function, and the maximum fitting error corresponding to each candidate independent variable interval is equal.

[0013] The third determining module is used to determine the target linear fitting coefficients corresponding to the target independent variable interval, and to process the communication signal in response to the service processing instruction based on the target linear fitting coefficients.

[0014] Thirdly, this application also provides a computer device, including a memory and a processor, the memory storing a computer program, the processor executing the computer program to implement the steps of the method described in the first aspect above.

[0015] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method described in the first aspect above.

[0016] Fifthly, this application also provides a computer program product, which includes a computer program that, when executed by a processor, implements the steps of the method described in the first aspect above.

[0017] The aforementioned business data processing method, apparatus, device, storage medium, and program product receive a business processing instruction to process a communication signal, and determine a target nonlinear function and the values ​​of its independent variables based on the instruction. This allows the determination of the target independent variable interval from multiple candidate independent variable intervals corresponding to the target nonlinear function. Furthermore, by determining the target linear fitting coefficient corresponding to the target independent variable interval, the communication signal is processed in response to the business processing instruction based on the target linear fitting coefficient, thus realizing the communication signal processing process. Since the multiple candidate independent variable intervals are obtained by non-uniformly dividing the range of independent variable values ​​for the target nonlinear function, a minimum number of independent variable intervals can be obtained. Consequently, the computational complexity and computation time required to determine the target linear fitting coefficient from the linear fitting coefficients corresponding to the multiple independent variable intervals are significantly reduced. When processing the communication signal based on the target fitting coefficient in response to the business processing instruction, the computational load is reduced while processing efficiency is improved. In addition, since the maximum fitting error corresponding to each candidate independent variable interval is equal, there will be no problem of excessive or poor maximum fitting error for a certain candidate independent variable interval, thus ensuring that the maximum fitting error meets the preset error requirements and improving the fitting accuracy. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating a business data processing method in one embodiment;

[0019] Figure 2 This is a flowchart illustrating the process of determining the candidate independent variable interval in one embodiment;

[0020] Figure 3 This is a schematic diagram of the non-uniform partitioning process in one embodiment;

[0021] Figure 4 This is a flowchart illustrating the interval determination operation in one embodiment;

[0022] Figure 5 This is a flowchart illustrating the process of adjusting the linear fitting coefficients in one embodiment;

[0023] Figure 6 This is a flowchart illustrating the process of determining the initial maximum fitting error in one embodiment;

[0024] Figure 7 This is a flowchart illustrating the response to a service processing instruction in one embodiment;

[0025] Figure 8 This is a logic diagram of non-uniform segmentation in one embodiment;

[0026] Figure 9This is a schematic diagram of the result of non-uniform segmentation in one embodiment;

[0027] Figure 10 This is a schematic diagram of the interval segmentation result with the zero point as the endpoint in one embodiment;

[0028] Figure 11 This is a schematic diagram of the interval segmentation result where the zero point is not an endpoint in one embodiment;

[0029] Figure 12 This is a schematic diagram of the segmentation result of uniform segmentation in one embodiment;

[0030] Figure 13 This is a schematic diagram of another non-uniform segmentation result in one embodiment;

[0031] Figure 14 This is a structural block diagram of a business data processing device in one embodiment;

[0032] Figure 15 This is a structural block diagram of the business data processing apparatus in another embodiment;

[0033] Figure 16 This is an internal structural diagram of a computer device in one embodiment;

[0034] Figure 17 This is a diagram of the internal structure of a computer device in another embodiment. Detailed Implementation

[0035] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0036] During operation, the terminal relies on chips within it to perform various calculations to achieve communication or data processing functions. For example, the chips in the terminal may include demodulation modules, decoding modules, radio frequency modules, or channel estimation modules. Each module may involve different nonlinear functions when processing communication signals. Due to hardware implementation limitations, the values ​​of nonlinear functions cannot be directly calculated and must be obtained through fitting.

[0037] In traditional techniques, for a target nonlinear function to be fitted, the range of values ​​for the independent variable of the target nonlinear function is evenly divided into multiple sub-intervals, and the fitting coefficients for each sub-interval are determined based on the maximum fitting error corresponding to each sub-interval. Since the slope of the nonlinear function varies at different values ​​of the independent variable, if the required fitting error is small, the shorter the length of each evenly divided interval, the more segmented intervals are obtained, and the more fitting coefficients are generated. Thus, during the chip's computation, the value of the target nonlinear function is determined based on the pre-determined fitting coefficients.

[0038] However, in traditional techniques, on the one hand, uniformly dividing the range of independent variables results in a large number of sub-intervals, leading to a large number of fitting coefficients. This increases the computational complexity, computational load, and runtime of obtaining these coefficients during chip processing, impacting fitting efficiency. Simultaneously, the terminal needs to store more fitting coefficients for each sub-interval, increasing storage costs. On the other hand, uniformly dividing the range of direct independent variables can lead to varying maximum fitting errors across sub-intervals. This may result in some sub-intervals having excessive or insufficient fitting accuracy, failing to fully match the maximum fitting error required by the chip in the terminal for processing communication signals or service data, thus resulting in lower computational accuracy.

[0039] In view of this, embodiments of this application provide a service data processing method. This method receives a service processing instruction instructing the processing of a communication signal, and determines a target nonlinear function and the values ​​of its independent variables based on the instruction. This allows the determination of the target independent variable interval from multiple candidate independent variable intervals corresponding to the target nonlinear function. Furthermore, by determining the target linear fitting coefficient corresponding to the target independent variable interval, the communication signal is processed in response to the service processing instruction based on the target linear fitting coefficient, thus realizing the processing of the communication signal. Since the multiple candidate independent variable intervals are obtained by non-uniformly dividing the range of independent variable values ​​of the target nonlinear function, a minimum number of independent variable intervals can be obtained. Consequently, the computational complexity and computation time required to determine the target linear fitting coefficient from the linear fitting coefficients corresponding to the multiple independent variable intervals are significantly reduced. When processing the communication signal based on the target fitting coefficient in response to the service processing instruction, the computational load is reduced while processing efficiency is improved. In addition, since the maximum fitting error corresponding to each candidate independent variable interval is equal, there will be no problem of excessive or poor maximum fitting error for a certain candidate independent variable interval, thus ensuring that the maximum fitting error meets the preset error requirements and improving the fitting accuracy.

[0040] like Figure 1The diagram shown is a flowchart illustrating a business data processing method provided in this application embodiment. This embodiment uses the application of this business data processing method to a terminal as an example for illustration. It is understood that this method can also be applied to a server, and further to a system including both a terminal and a server, and is implemented through interaction between the terminal and the server. The terminal can be, but is not limited to, various personal computers, laptops, smartphones, tablets, IoT devices, and portable wearable devices. IoT devices can include smart speakers, smart TVs, smart air conditioners, smart vehicle devices, etc. Portable wearable devices can include smartwatches, smart bracelets, head-mounted devices, etc. The server can be implemented using a standalone server or a server cluster composed of multiple servers. In this application embodiment, the business data processing method includes the following steps:

[0041] Step 101: Receive a service processing instruction, determine the target nonlinear function and the value of the independent variable of the target nonlinear function according to the service processing instruction, and the service processing instruction is used to instruct the processing of the communication signal.

[0042] During operation, the terminal can receive service processing instructions and process communication signals according to these instructions. Optionally, the communication signals can be various signals such as noise signals, audio signals, and radio frequency signals that the terminal needs to process, or signals containing other communication data such as channel capacity, measurement data, or decoding data. Since the functions that the terminal can implement are quite complex, this application embodiment will not provide a complete list. Based on this, the service processing instructions can be used to instruct the terminal to process the communication signals using appropriate signal processing methods to achieve the corresponding functions within the terminal. The terminal typically contains a chip; therefore, it is the chip within the terminal that receives the service processing instructions to achieve the corresponding functions. Specifically, the chip in the terminal can determine the target nonlinear function related to the service processing instructions and the value of the independent variable of the target nonlinear function according to the service processing instructions, thereby further determining the fitted value of the target nonlinear function at the value of the independent variable. Optionally, the target nonlinear function can be a power function, reciprocal function, exponential function, square root function, Bessel function, trigonometric function, or logarithmic function, etc. Of course, the processing of communication signals in the terminal is quite complex, and it may also involve the use of other nonlinear functions. Therefore, the target nonlinear function can also be other nonlinear functions, which will not be specifically limited or elaborated here. For example, the chip in the terminal may include multiple modules such as a demodulation module, a decoding module, an RF module, or a channel estimation module. The channel estimation module often involves reciprocal functions or square root functions in its calculations, while the RF module often involves exponential functions or logarithmic functions in its calculations.

[0043] Step 102: Determine the target independent variable interval from multiple candidate independent variable intervals corresponding to the target nonlinear function; wherein, the multiple candidate independent variable intervals are obtained by non-uniformly dividing the range of independent variable values ​​of the target nonlinear function, and the maximum fitting error corresponding to each candidate independent variable interval is equal.

[0044] The terminal can store multiple candidate independent variable intervals corresponding to the target nonlinear function. Each candidate independent variable interval is pre-determined based on a non-uniform partitioning of the independent variable range of the target nonlinear function. For example, for log(x), the range of the independent variable x is x>0. Therefore, when the target nonlinear function is log(x), the range of the independent variable is x>0. The range of x>0 is then non-uniformly partitioned to obtain multiple candidate independent variable intervals corresponding to the target nonlinear function. Non-uniform partitioning refers to sequentially dividing the range of the independent variable into multiple candidate independent variable intervals of unequal length.

[0045] Furthermore, when obtaining candidate independent variable intervals through non-uniform partitioning, the maximum fitting error corresponding to each candidate independent variable interval is equal. The maximum fitting error refers to the maximum error between the fitted value calculated using the linear fitting coefficients and independent variable values ​​corresponding to the candidate independent variable interval, and the theoretical value calculated using the target nonlinear function. This maximum fitting error can refer to either relative or absolute fitting error. Optionally, due to the influence of hardware devices in the chip, the maximum fitting error corresponding to each candidate independent variable interval should be less than the upper limit of the maximum fitting error during the operation of each module in the terminal chip.

[0046] Once the value of the independent variable of the target nonlinear function is determined, the target independent variable interval is selected from multiple candidate independent variable intervals corresponding to the target nonlinear function. Based on this target independent variable interval, the communication signal is processed in response to the service processing instruction. For example, if the value of the independent variable is 0.3, and one of the candidate independent variable intervals has a range of [0.1, 0.4], then this candidate independent variable interval with a range of [0.1, 0.4] is selected as the target independent variable interval.

[0047] Step 103: Determine the target linear fitting coefficient corresponding to the target independent variable interval, and process the communication signal in response to the service processing instruction based on the target linear fitting coefficient.

[0048] Specifically, in addition to storing multiple candidate independent variable intervals, the terminal also stores the linear fitting function corresponding to each candidate independent variable interval. Optionally, the terminal stores a coefficient correspondence table, which includes the correspondence between multiple candidate independent variable intervals and linear fitting coefficients.

[0049] After determining the target independent variable interval, the target linear fitting coefficient corresponding to the target independent variable interval is determined based on the linear fitting coefficients corresponding to each candidate independent variable interval. That is, the linear fitting coefficients associated with the candidate independent variable intervals corresponding to the target independent variable interval are used as the target linear fitting coefficients, and the communication signal is processed in response to the service processing instruction based on the target linear fitting coefficients.

[0050] Optionally, the linear fitting coefficients corresponding to each candidate independent variable interval include intercept fitting coefficients and slope fitting coefficients. Based on these intercept fitting coefficients and slope fitting coefficients, a preset linear fitting function can be determined. Thus, given the value of the independent variable, the fitted value of the target nonlinear function at the value of the independent variable is determined in the form of the preset linear fitting function, based on these intercept fitting coefficients and slope fitting coefficients. Based on this fitted value, the communication signal is processed in response to the service processing instruction. Optionally, the preset linear fitting function can be y = Ax + B, where A is the intercept fitting coefficient, B is the slope fitting coefficient, x is the value of the independent variable, and y is the obtained fitted value; or, the preset linear fitting function can be y = A(x - x1) + B, where A is the intercept fitting coefficient, B is the slope fitting coefficient, x is the value of the independent variable, x1 is the reference independent variable value, and y is the obtained fitted value.

[0051] Of course, in one feasible approach, after determining the target nonlinear function, the terminal can also calculate online multiple candidate independent variable intervals and the corresponding linear fitting coefficients for each independent variable interval, and further determine the target independent variable interval and the target linear fitting coefficient, so as to determine the fitting value of the target nonlinear function at the independent variable value based on the target linear fitting coefficient, and process the communication signal in response to the service processing instruction based on the fitting value.

[0052] The aforementioned business data processing method receives a business processing instruction to process a communication signal, and determines a target nonlinear function and the values ​​of its independent variables based on the instruction. This allows the determination of the target independent variable interval from multiple candidate independent variable intervals corresponding to the target nonlinear function. Furthermore, by determining the target linear fitting coefficient corresponding to the target independent variable interval, the communication signal is processed in response to the business processing instruction based on the target linear fitting coefficient, thus realizing the communication signal processing process. Since the multiple candidate independent variable intervals are obtained by non-uniformly dividing the range of independent variable values ​​for the target nonlinear function, this embodiment of the application accurately divides the range of independent variable values ​​with the fewest possible intervals, based on the slope of the target nonlinear function at different independent variable values. Consequently, compared to existing technologies, the number of fitting coefficients is also reduced. Therefore, by obtaining the minimum number of independent variable intervals, the maximum fitting error corresponding to each interval is also equal. Consequently, the computational complexity required to determine the target linear fitting coefficient from the linear fitting coefficients corresponding to multiple independent variable intervals is reduced linearly, and the computation time is decreased. When processing communication signals in response to the service processing instruction based on the target fitting coefficient, the processing efficiency is improved while reducing the amount of computation. In addition, since the maximum fitting error corresponding to each candidate independent variable interval is equal, there will be no problem of excessive or poor maximum fitting error for any candidate independent variable interval, thus ensuring that the maximum fitting error meets the preset error requirements and improving the fitting accuracy.

[0053] In one embodiment, such as Figure 2 The diagram illustrates a flowchart of a method for determining candidate independent variable intervals according to an embodiment of this application. Before determining the target independent variable interval where the value of the independent variable lies from multiple candidate independent variable intervals corresponding to the target nonlinear function, the method further includes:

[0054] Step 201: Based on the initial maximum fitting error and the target nonlinear function, iteratively perform multiple non-uniform partitioning processes on the range of values ​​of the independent variable until the iteration stopping condition is met.

[0055] The range of values ​​for the independent variable includes the interval between the first endpoint and the second endpoint. The iteration stops when the first endpoint is the endpoint of the first independent variable interval among the multiple independent variable intervals, and the second endpoint is the endpoint of the last independent variable interval among the multiple independent variable intervals.

[0056] Specifically, the range of values ​​for the independent variable corresponding to the target nonlinear function can be determined based on the target nonlinear function, and a preset initial maximum fitting error can be obtained. Based on the initial maximum fitting error, the range of values ​​for the independent variable is divided into multiple non-uniform intervals. The fitting coefficients corresponding to each interval are then determined until the iteration conditions are met, thereby obtaining multiple candidate intervals of independent variables with equal maximum fitting errors and the fitting coefficients corresponding to each candidate interval.

[0057] Specifically, for example, if the range of the independent variable is represented as [a, b], then a is the first endpoint, b is the second endpoint, and the range of the independent variable refers to the range between a and b. The range of the independent variable can be non-uniformly divided sequentially to obtain multiple intervals. Optionally, the non-uniform division can be performed progressively from the first endpoint in ascending order until the second endpoint is reached, or it can be performed from the second endpoint in descending order until the first endpoint is reached. The following text uses the example of progressively dividing the range of the independent variable in ascending order from the first endpoint. Similarly, the process of dividing the range of the independent variable in descending order from the second endpoint is the same as the ascending process and will not be repeated.

[0058] For example, for a single non-uniform partitioning process, if the range of the independent variable is [0.5, 1), the resulting multiple independent variable intervals can be [0.5, 0.65), [0.65, 0.68), [0.68, 0.9), and [0.9, 1). Among them, [0.5, 0.65) is the first independent variable interval, and [0.9, 1) is the last independent variable interval. During the non-uniform partitioning process, there may be instances where the partitioning is performed with the endpoint of the last independent variable interval being greater than the second endpoint. For example, the partitioning might be performed with the last independent variable interval set to [0.9, 1.2), but the actual last independent variable interval is [0.9, 1). This discrepancy between the partitioning and the actual interval can result in the maximum fitting error of the last independent variable interval being less than the initial maximum fitting error. Therefore, it is necessary to determine an updated maximum fitting error based on the initial maximum fitting error, and then perform the non-uniform partitioning process again based on this updated maximum fitting error, until the first endpoint is the endpoint of the first independent variable interval among the multiple partitioned intervals, and the second endpoint is the endpoint of the last independent variable interval among the multiple partitioned intervals.

[0059] It should be noted that the above example uses a half-open interval that is closed on the left and open on the right. Of course, the range of values ​​for the independent variable can be an open interval, a closed interval, or a half-open interval that is open on the left and closed on the right. Correspondingly, the interval of the candidate independent variable can also be an open interval, a closed interval, or a half-open interval. This application does not limit this, and the description of the interval range in the following text is also for example and not a limitation.

[0060] Step 202: The multiple independent variable intervals obtained by non-uniformly dividing the range of values ​​of the independent variable at the time of stopping iteration are taken as the multiple candidate independent variable intervals, and the linear fitting coefficients corresponding to each independent variable interval at the time of stopping iteration are taken as the linear fitting coefficients corresponding to each candidate independent variable interval.

[0061] When the iteration update condition is met, the iteration stops. The multiple independent variable intervals obtained by non-uniformly dividing the range of values ​​of the independent variable at the time of stopping iteration are then used as candidate independent variable intervals. Simultaneously, the linear fitting coefficients corresponding to each independent variable interval at the time of stopping iteration are used as the linear fitting coefficients corresponding to each candidate independent variable interval. At this point, the first endpoint is the endpoint of the first candidate independent variable interval, the second endpoint is the endpoint of the last candidate independent variable interval, and the maximum fitting error of each candidate independent variable interval is equal.

[0062] In this embodiment, based on the initial maximum fitting error and the target nonlinear function, the range of values ​​for the independent variable is iteratively divided multiple times into non-uniform intervals. This fully considers the possibility of excessive maximum fitting error in the last interval of the independent variable, further reducing the maximum fitting error of each candidate interval. Thus, while achieving accurate non-uniform division of the range of values ​​for the independent variable, the maximum fitting error of each candidate interval is also equal, preventing any individual candidate interval from having excessive or poor maximum fitting error. This ensures the accuracy and consistency of the fitted value of the target nonlinear function, effectively improving the accuracy of subsequent response business processing instructions.

[0063] As mentioned above, multiple iterations of non-uniform partitioning are required to determine the intervals of the multiple candidate independent variables and the corresponding fitting coefficients. The process of this single non-uniform partitioning will be explained below.

[0064] Please refer to Figure 3 This document illustrates a flowchart of a non-uniform partitioning process provided in an embodiment of this application. The i-th non-uniform partitioning process in this multi-stage non-uniform partitioning process includes:

[0065] Step 301: Determine the target maximum fitting error corresponding to the i-th non-uniform partitioning process based on the initial maximum fitting error. Specifically, when i = 1, the target maximum fitting error corresponding to the i-th non-uniform partitioning process is the initial maximum fitting error; when i > 1, the target maximum fitting error corresponding to the i-th non-uniform partitioning process is determined based on the target maximum fitting error corresponding to the (i-1)-th non-uniform partitioning process.

[0066] Specifically, as mentioned above, the initial maximum fitting error is a preset maximum fitting error. Optionally, it can be determined based on the properties or curve image of the target nonlinear function. The terminal can obtain the initial maximum fitting error and perform the first non-uniform partitioning process based on the initial maximum fitting error. If the first non-uniform partitioning process does not meet the iteration stopping condition, the updated maximum fitting error is determined based on the initial maximum fitting error, which is the aforementioned target maximum fitting error. Thus, the next non-uniform partitioning process is performed based on the target maximum fitting error.

[0067] The target maximum fitting error refers to the maximum fitting error corresponding to each non-uniform partitioning process. When i = 1, i.e., for the first non-uniform partitioning process, this initial maximum fitting error is directly used as the target maximum fitting error. When i > 1, i.e., for non-first non-uniform partitioning processes, the target maximum fitting error for this non-uniform partitioning process needs to be determined based on the target maximum fitting error corresponding to the previous non-uniform partitioning process. The method for determining the target maximum fitting error for non-first non-uniform partitioning processes will be explained below.

[0068] In one embodiment, when i is greater than 1, the process for determining the maximum fitting error of the target corresponding to the i-th non-uniform partitioning process includes:

[0069] Method 1: Subtract the target maximum fitting error corresponding to the (i-1)th non-uniform partitioning process from the preset step size to obtain the target maximum fitting error corresponding to the i-th non-uniform partitioning process.

[0070] Optionally, the terminal may store a preset step size Δ, and the maximum fitting error of the target corresponding to the (i-1)th non-uniform partitioning process is denoted as... Let the maximum fitting error corresponding to the i-th non-uniform partitioning process be denoted as . Then it can be based on Determine the maximum fitting error corresponding to the i-th non-uniform partitioning process. The value of the preset step size can be determined based on the actual situation or the properties of the target nonlinear function, and is not specifically limited here.

[0071] Method 2: Obtain the maximum fitting error of the interval corresponding to the last independent variable interval, and take the average of the maximum fitting error of the interval and the target maximum fitting error corresponding to the (i-1)th non-uniform partitioning process as the target maximum fitting error corresponding to the i-th non-uniform partitioning process.

[0072] Optionally, as mentioned above, if the iteration stopping condition is not met, the maximum fitting error of the last independent variable interval after the non-uniform partitioning is less than the maximum fitting error of other independent variable intervals, that is, it is not equal to the target maximum fitting error. Therefore, the maximum fitting error of the interval corresponding to the last independent variable interval obtained by the (i-1)th non-uniform partitioning can be determined. and through Determine the target maximum fitting error corresponding to the i-th non-uniform partitioning process.

[0073] Method 3: The weighted average of the maximum fitting errors corresponding to the multiple independent variable intervals obtained by the (i-1)th non-uniform partitioning process is taken as the target maximum fitting error corresponding to the i-th non-uniform partitioning process.

[0074] Optionally, the maximum fitting error corresponding to multiple independent variable intervals obtained from the (i-1)th non-uniform partitioning process and the weighting coefficients corresponding to each independent variable interval can be obtained. The target maximum fitting error corresponding to the i-th non-uniform partitioning process can be obtained by calculating the weighted average of the maximum fitting errors corresponding to each independent variable interval. Specifically, through... Determine the maximum fitting error of the target corresponding to the i-th non-uniform partitioning process, satisfying... Among them, w n The weighting coefficients are the values ​​corresponding to the nth independent variable interval in the (i-1)th non-uniform partitioning process. is the fitting coefficient corresponding to the nth independent variable interval in the (i-1)th non-uniform partitioning process.

[0075] It should be noted that other methods can also be used to determine the target maximum fitting error corresponding to the i-th non-uniform partitioning process. This application does not specifically limit this method, as long as the target maximum fitting error corresponding to the i-th non-uniform partitioning process is less than the target maximum fitting error corresponding to the (i-1)-th non-uniform partitioning process.

[0076] Step 302: Based on the maximum fitting error of the target corresponding to the i-th non-uniform partitioning process, perform non-uniform partitioning on the range of values ​​of the independent variable.

[0077] In the i-th non-uniform partitioning process, after determining the corresponding target maximum fitting error, the range of independent variable values ​​can be non-uniformly partitioned based on this target maximum fitting error, thus obtaining multiple independent variable intervals. The following section will explain the process of performing non-uniform partitioning of the independent variable value range to determine the endpoints of each interval and the fitting coefficients.

[0078] In one embodiment, the range of values ​​of the independent variable is non-uniformly divided according to the target maximum fitting error corresponding to the i-th non-uniform division process, including: performing multiple interval determination operations on the range of values ​​of the independent variable according to the target maximum fitting error corresponding to the i-th non-uniform division process to obtain multiple independent variable intervals corresponding to the i-th non-uniform division process, wherein each independent variable interval includes the interval range between the endpoints of the first interval and the endpoints of the second interval.

[0079] Similar to the range of values ​​of the independent variable between the first and second endpoints, the range of the independent variable includes the range between the first and second endpoints, and the range of each independent variable range can be represented by the corresponding first and second endpoints.

[0080] Please refer to Figure 4 The diagram illustrates a flowchart of an interval determination operation provided in an embodiment of this application. The j-th interval determination operation in a series of interval determination operations includes:

[0081] Step 401: When j=1, take the first endpoint as the first interval endpoint of the j-th independent variable interval.

[0082] Step 402: When j>1, take the second interval endpoint of the (j-1)th independent variable interval as the first interval endpoint of the jth independent variable interval.

[0083] Step 403: Determine the second interval endpoints and corresponding linear fitting coefficients of the j-th independent variable interval based on the target maximum fitting error corresponding to the i-th non-uniform partitioning process.

[0084] Specifically, performing the j-th interval determination operation yields the first and second interval endpoints of the j-th independent variable interval, thereby determining the j-th independent variable interval based on the first and second interval endpoints.

[0085] Specifically, for each independent variable interval, the endpoints of the second interval and the corresponding linear fitting function can be determined based on the target maximum fitting error corresponding to the i-th non-uniform partitioning process.

[0086] The determination of the first interval endpoint of each independent variable interval is different. Specifically, when j=1, that is, for the first independent variable interval, the first endpoint included in the range of values ​​of the independent variable is directly taken as the first interval endpoint of the first independent variable interval; when j>1, the second interval endpoint of the (j-1)th independent variable interval is directly taken as the first interval endpoint of the jth independent variable interval.

[0087] In this embodiment of the application, by performing non-uniform partitioning processing, the number of independent variable intervals obtained will not be excessive, and correspondingly, the number of fitting coefficients will not be excessive. The storage cost required by the terminal is reduced. When determining the target independent variable interval and the target linear fitting coefficient, the complexity of the terminal comparison calculation is reduced linearly, the running time is reduced, and the computing efficiency is improved, so as to improve the response speed to business processing instructions.

[0088] Specifically, the process of determining the second interval endpoints and corresponding linear fitting coefficients of the j-th independent variable interval in the i-th non-uniform partitioning process will be explained below.

[0089] In one embodiment, determining the second interval endpoint and corresponding linear fitting coefficient of the j-th independent variable interval based on the target maximum fitting error corresponding to the i-th non-uniform partitioning process includes: constructing a system of equations based on a pre-set first constraint, second constraint, and third constraint, and solving the system of equations to obtain the second interval endpoint and corresponding linear fitting coefficient of the j-th independent variable interval.

[0090] The first constraint condition includes that the fitting error of the first interval endpoint of the j-th independent variable interval is equal to the target maximum fitting error; the second constraint condition includes that the fitting error of one and only one point other than the interval endpoint in the j-th independent variable interval is equal to the target maximum fitting error; and the third constraint condition includes that the fitting error of the second interval endpoint of the j-th independent variable interval is equal to the target maximum fitting error.

[0091] Specifically, as mentioned above, the preset linear fitting function can be of the form y = Ax + B, or y = A(x - x1) + B. Taking the form y = A(x - x1) + B as an example, the preset linear fitting function... Recorded as:

[0092]

[0093] Among them, A j and Let be the fitting coefficient for the j-th independent variable interval. Let x be the endpoint of the first interval of the j-th independent variable interval, and let x be the value of the independent variable.

[0094] For any value of the independent variable within the range of independent variables, its corresponding maximum fitting error can be obtained through a preset fitting error function e1(x). Optionally, this preset fitting error function can be a relative fitting error function or an absolute fitting error function. When the preset fitting error function is a relative fitting error function, the preset fitting error function is:

[0095]

[0096] Where f(x) is the objective nonlinear function; sign(·) is the sign function. For sign(-f"(x)), when f(x) is a concave function, sign(-f"(x)) takes the value of -1, and when f(x) is a convex function, sign(-f"(x)) takes the value of 1.

[0097] Let ε be the maximum fitting error corresponding to the i-th non-uniform partitioning process. * Let the endpoint of the first interval of the j-th independent variable be denoted as . The first constraint condition includes the fact that the fitting error of the first interval endpoint of the j-th independent variable interval is equal to the target maximum fitting error, that is:

[0098]

[0099] Accordingly, the point in the j-th independent variable interval where, excluding the endpoints, the fitting error is exactly equal to the maximum fitting error of the target variable is denoted as . The second constraint condition includes the requirement that the fitting error of one and only one point in the j-th independent variable interval, excluding the endpoints of the interval, equals the target maximum fitting error, i.e.:

[0100]

[0101] At the same time, since there is only one The fitting error is equal to ε * Therefore, the preset fitting error function e1(x) is in To reach the maximum value, i.e.

[0102]

[0103] Accordingly, the endpoint of the second interval of the j-th independent variable interval is denoted as... The third constraint condition includes the fact that the fitting error of the endpoints of the second interval of the j-th independent variable interval is equal to the target maximum fitting error, that is:

[0104]

[0105] Therefore, based on the first, second, and third constraints, a system of equations can be constructed by simultaneously applying equations (1) to (6). By solving this system of equations, the endpoint of the second interval of the j-th independent variable interval can be obtained. and the corresponding linear fitting coefficients A j and Specifically, in solving the problem, the terminal can obtain the solution based on equations (1) to (3). The terminal can be obtained based on equations (2), (4), and (5). And A jThe terminal can be obtained based on equations (2) and (6). Alternatively, it can be determined through mathematical numerical solutions. A j as well as For example, the method of finding the mathematical numerical solution can be implemented using the vpasolve function in the software MATLAB. The value obtained by the search is highly accurate and can be regarded as the theoretical value.

[0106] As mentioned above, the preset linear fitting function can also be of the form y = Ax + B. In this case, the preset linear fitting function... Recorded as:

[0107]

[0108] satisfy:

[0109]

[0110] Therefore, after obtaining A j , as well as Based on this, B can also be determined based on the form of equation (8). j The value of A, to make A j and B j As the fitting coefficient for the j-th independent variable interval.

[0111] Therefore, in one feasible approach, the terminal can store A corresponding to each candidate independent variable interval. j and This is used to determine the fitted value of the target nonlinear function; in another possible implementation, the terminal can store the corresponding A values ​​for each candidate independent variable interval. j and B j This is used to determine the fitted value of the target nonlinear function.

[0112] As mentioned above, the preset fitting error function can also be an absolute fitting error function, that is:

[0113]

[0114] The fitting coefficients are also determined based on this absolute fitting error function. The determination process is the same as that for the relative fitting error function, and will not be repeated here.

[0115] In this embodiment, the above method allows for the determination of the first and second endpoints of each independent variable interval obtained from non-uniform segmentation, along with their corresponding fitting coefficients, based on the initial maximum fitting error. Simultaneously, considering the possibility that the second endpoint of the last calculated independent variable interval might exceed the range of independent variable values, and to avoid overfitting of the last independent variable interval while the fitting performance of other independent variable intervals just matches the upper limit of the maximum fitting error, the target maximum fitting error is updated. Consequently, the segmented interval length of the last independent variable interval increases, while the segmented interval lengths of other independent variable intervals decrease accordingly. This results in a corresponding reduction in the maximum fitting error, ensuring that the maximum fitting errors for all independent variable intervals are equal and achieving a lower upper limit for the maximum fitting error.

[0116] The process of determining multiple independent variable intervals and corresponding fitting coefficients described above applies to target nonlinear functions where the range of independent variable values ​​does not include zeros. However, business processing instructions often involve multiple target nonlinear functions where the range of independent variable values ​​includes zeros. For example, when fitting f(x) = log2(x) to x within the range [0.5, 1), f(x) ≈ 0 when x is close to 1, which greatly amplifies the maximum fitting error, resulting in a large maximum fitting error in the interval of independent variable values ​​close to x = 1, failing to meet the requirement of being less than the preset maximum fitting error upper limit. As another example, when fitting f(x) = sin(x) to x within the range [0, 1), f(x) = 0 when x = 0, making the relative fitting error function e1(x) unusable.

[0117] Based on this, for a target nonlinear function whose independent variable range includes a zero point, the zero point is set to a pre-defined linear fitting function. The intersection point with the target nonlinear function f(x) results in zero absolute or relative error at the zero point. The optimization principle involves non-uniformly dividing the range of values ​​for the independent variable of the target nonlinear function, resulting in multiple independent variable intervals. The following section will explain the process of determining the fitting coefficients corresponding to the characteristic independent variable intervals when the characteristic independent variable intervals contain the zeros of the target nonlinear function.

[0118] In one embodiment, such as Figure 5 The diagram illustrates a flowchart of adjusting linear fitting coefficients according to an embodiment of this application. The linear fitting coefficients corresponding to each independent variable interval at the time of stopping iteration are used as the linear fitting coefficients corresponding to each candidate independent variable interval, including:

[0119] Step 501: Determine whether there exists a characteristic independent variable interval among the multiple independent variable intervals, where the characteristic independent variable interval contains the zeros of the target nonlinear function.

[0120] Step 502: If a characteristic independent variable interval exists, the linear fitting coefficients corresponding to the characteristic independent variable interval are adjusted, and the adjusted linear fitting coefficients are used as the linear fitting coefficients of the corresponding candidate independent variable intervals.

[0121] For the target nonlinear function whose independent variable range includes zeros, the interval containing zeros in the non-uniform partitioning process is taken as the characteristic independent variable interval. Zeros may be endpoints of this characteristic independent variable interval or lie within it. In cases where zeros exist, the method for determining the linear fitting coefficients of the characteristic independent variable interval needs to be updated. This process will be explained below.

[0122] In one embodiment, adjusting the linear fitting coefficients corresponding to the feature variable interval includes: recalculating the linear fitting coefficients corresponding to the feature variable interval based on the target constraint when the endpoints of the first or second interval of the feature variable interval are zero; the target constraint includes that the fitting error of the endpoints of the first interval of the feature variable interval is equal to 0.

[0123] Specifically, for a feature variable interval containing a zero, the above equations can be adjusted based on the target constraints, depending on whether the zero is the first or second interval endpoint of the feature variable interval, and the linear fitting coefficients corresponding to the feature variable interval can be obtained based on the adjusted equations.

[0124] Taking the interval of the characteristic independent variable as the j-th independent variable interval, the objective constraint includes the first interval endpoint of the characteristic independent variable interval. The fitting error is equal to 0, that is:

[0125]

[0126] Therefore, under the condition that other conditions remain unchanged, replacing equation (3) with equation (10) yields a new set of equations. Solving the new set of equations yields a more accurate fitting coefficient corresponding to the interval of the characteristic independent variables. At this time, the obtained It is 0.

[0127] In one embodiment, adjusting the linear fitting coefficients corresponding to the feature variable interval includes: when the feature variable interval contains a zero point and the zero point is not an interval endpoint of the feature variable interval, recalculating the linear fitting coefficients of the feature variable interval based on the zero point, the first interval endpoint of the feature variable interval, and the value of the target nonlinear fitting function at the zero point.

[0128] When the zero point is within the interval of the characteristic independent variable but is not the first or second interval endpoint of the characteristic independent variable interval, a new set of equations is constructed based on the zero point, the first interval endpoint of the characteristic independent variable interval, and the value of the target nonlinear fitting function at the zero point, and the linear fitting coefficients corresponding to the characteristic independent variable interval are obtained through the new set of equations.

[0129] Continuing with the characteristic variable interval as the j-th independent variable interval, specifically, the fitting coefficient A corresponding to the characteristic variable interval... j It can be determined by the following formula:

[0130]

[0131] Where x0 is the zero point, and f(x0) is the value of the objective nonlinear function at the zero point. The endpoint of the first interval of the characteristic independent variable interval, This represents the values ​​of the preset linear fitting function at the endpoints of the first interval of the characteristic independent variable interval. Clearly, due to the sequential division, the points on the preset linear fitting function... Since it is known, the fitting coefficient A corresponding to the interval of the characteristic independent variable can be obtained based on equation (11). j Therefore, when the zero point is in the interval of the characteristic independent variable and is not the first or second interval endpoint of the characteristic independent variable interval, a new set of equations is constructed according to equations (1) to (3), (6) and (11). By solving the new set of equations, the fitting coefficients corresponding to the characteristic independent variable interval and the second interval endpoint are obtained. The obtained fitting coefficients are more accurate fitting coefficients.

[0132] In this embodiment of the application, the solution process is optimized for cases where the range of independent variable values ​​includes zero points, so as to avoid affecting the accuracy of the fitting coefficient of the independent variable range due to the presence of zero points.

[0133] As mentioned above, the terminal can perform multiple non-uniform partitioning processes on the range of values ​​of the independent variable based on the initial maximum fitting error and the target nonlinear function. Therefore, a suitable initial maximum fitting error plays an important role in the non-uniform partitioning process. The process of determining the initial maximum fitting error will be explained below.

[0134] Please refer to Figure 6 This document illustrates a flowchart of a method for determining the initial maximum fitting error, as provided in an embodiment of this application. The process for determining the initial maximum fitting error includes:

[0135] Step 601: Obtain the candidate maximum fitting error and determine whether the candidate maximum fitting error meets the preset conditions.

[0136] Step 602: If the preset conditions are met, the candidate maximum fitting error is taken as the initial maximum fitting error.

[0137] Step 603: If the preset conditions are not met, the candidate maximum fitting error is adjusted until the adjusted candidate maximum fitting error meets the preset conditions.

[0138] The preset conditions include: the number of multiple reference independent variable intervals obtained by non-uniformly dividing the range of independent variable values ​​is equal to the preset target number, and the target number is an integer multiple of the preset value.

[0139] Specifically, for chips or other hardware devices in a terminal, a target number N for the number of independent variable intervals is more conducive to the implementation of the chip and various hardware components. Optionally, this target number N needs to be an integer multiple of a preset value, where the preset value can be 4. In other words, for each target nonlinear function, the required target number N can be predetermined. If the number of independent variable intervals obtained after a candidate maximum fitting error and non-uniform division of the independent variable value range based on the candidate maximum fitting error is not equal to the target number, then the candidate maximum fitting error should be adjusted, and non-uniform division should be performed based on the new candidate maximum fitting error until the number of independent variable intervals obtained is equal to the target number N. Then, the latest candidate maximum fitting error is used as the initial maximum fitting error.

[0140] Therefore, the preset conditions include ensuring that the number of reference independent variable intervals obtained by non-uniformly dividing the range of independent variable values ​​is equal to a preset target number, which is an integer multiple of the preset value. The terminal can obtain the candidate maximum fitting error and determine whether the candidate maximum fitting error meets the preset conditions. If the preset conditions are met, the candidate maximum fitting error is used as the initial maximum fitting error; if the preset conditions are not met, the candidate maximum fitting error is adjusted until the adjusted candidate maximum fitting error meets the preset conditions.

[0141] The adjustment process for the candidate maximum fitting error includes: if the number of multiple reference independent variable intervals obtained by non-uniformly dividing the range of independent variable values ​​is greater than the target number, the candidate maximum fitting error is increased; if the number of multiple reference independent variable intervals obtained by non-uniformly dividing the range of independent variable values ​​is less than the target number, the candidate maximum fitting error is decreased.

[0142] Specifically, let N be the number of reference variable intervals obtained by non-uniform partitioning based on the candidate maximum fitting error, and let N0 be the target number. That is, N0 is less than N. This indicates rounding down. In this case, the value of the candidate maximum fitting error should be increased. Optionally, a new candidate maximum fitting error can be obtained by adding a preset step size to the current candidate maximum fitting error. If N0 is greater than N, This indicates rounding up. In this case, the value of the candidate maximum fitting error should be reduced. Optionally, a new candidate maximum fitting error can be obtained by subtracting the preset step size from the current candidate maximum fitting error.

[0143] In this embodiment, by determining a more suitable initial maximum fitting error, the rationality and accuracy of the subsequent non-uniform partitioning process based on the initial maximum fitting error are improved, and the number of obtained independent variable intervals is equal to the preset required number of independent variable intervals. It should be noted that for the same target nonlinear function, different modules of the chip in the terminal may require different numbers of independent variable intervals and different maximum fitting errors when processing the target nonlinear function.

[0144] Based on this, upon receiving a service processing instruction, the target nonlinear function and the values ​​of its independent variables can be determined according to the instruction. After determining the target linear fitting coefficients, the communication signal can be processed in response to the service processing instruction based on the target linear fitting coefficients.

[0145] Please refer to Figure 7 This illustration shows a flowchart of a response service processing instruction provided in an embodiment of this application. The process of processing communication signals according to the target linear fitting coefficients in response to the service processing instruction includes:

[0146] Step 701: Determine the linear fitting function based on the target linear fitting coefficients.

[0147] Step 702: Process the communication signal using a linear fitting function in response to the service processing command.

[0148] As mentioned above, the target linear fitting coefficients can be determined based on the target independent variable interval where the independent variable values ​​are located, thereby obtaining the linear fitting function. Given the independent variable values, the fitted value of the target nonlinear function at the independent variable values ​​can be obtained based on the independent variable values ​​and the linear fitting function. Based on this fitted value, the business processing instructions can be responded to to realize the processing of communication signals.

[0149] In one embodiment, the chip in the terminal can receive service processing instructions during operation and process communication signals according to these instructions. Optionally, the communication signal can be various signals such as noise signals, audio signals, and radio frequency signals that the terminal needs to process, or a signal containing communication data such as channel capacity, measurement data, or decoding data. Based on this, the service processing instructions can be used to instruct the terminal to process the communication signals to achieve the corresponding functions in the terminal. Specifically, the chip in the terminal can include multiple modules such as a demodulation module, a decoding module, a radio frequency module, or a channel estimation module. The channel estimation module often involves reciprocal functions or square root functions during calculations, while the radio frequency module often involves exponential functions or logarithmic functions during calculations.

[0150] The chip in the terminal can determine the target nonlinear function and the value of the independent variable of the target nonlinear function according to the service processing instruction, and further determine the fitted value of the target nonlinear function at the value of the independent variable, thereby responding to the service processing instruction to process communication signals or communication data. In this embodiment, a method for fitting a nonlinear function is provided. By non-uniformly segmenting the nonlinear function, multiple intervals and corresponding fitting coefficients are obtained. After receiving the service processing instruction, the value of the independent variable of the target nonlinear function corresponding to the service processing instruction is determined. Based on the fitting coefficients corresponding to the interval where the independent variable value is located, the fitted value of the target nonlinear function is determined using the fitted linear function to further respond to the service processing instruction.

[0151] The terminal can obtain the maximum fitting error upper limit ε0, the target nonlinear function f(x), and the range of the independent variable x of f(x) from x to [a, b]. Through non-uniform partitioning, N segmented intervals are obtained, as shown in the table below:

[0152]

[0153] like Figure 8 As shown, it illustrates a logic diagram of non-uniform segmentation provided in an embodiment of this application. The following is based on... Figure 8 The process of obtaining each segment interval and fitting coefficient by non-uniformly segmenting the target nonlinear function is explained.

[0154] The preset maximum fitting error is denoted as ε. * The initial value of the preset maximum fitting error is ε0. The method for determining the starting point of each segment interval is as follows: if n = 1, then the starting point of the first segment interval is... If n ≥ 2, then the starting point of the nth interval is the ending point of the (n-1)th interval. The method for determining the endpoints of each segmented interval and the fitting coefficients is as follows: Based on the first constraint condition that the fitting error of the fitting function at the starting point is equal to the preset maximum fitting error, the fitting coefficients B are obtained. n Based on the second constraint that the fitting error of the fitting function at one and only one point between the starting and ending points is equal to the preset maximum fitting error, the fitting coefficient A is obtained. n Based on the third constraint that the fitting error of the fitting function at the endpoint equals the preset maximum fitting error, the endpoint is obtained.

[0155] Specifically, taking the first segmented interval as an example, let the fitting error function be denoted as e1(x). For the first constraint, the fitting function... The fitting error at the starting point is equal to ε. * ,Right now

[0156]

[0157]

[0158]

[0159] Based on the above formula, we can obtain:

[0160]

[0161] Therefore, it can be obtained The possible values ​​of:

[0162]

[0163] Based on the second constraint, the fitting function There is one and only one point The fitting error is ε * And the fitting error function is The maximum value is reached, that is, the fitting error function reaches its maximum value at... The first derivative of is equal to 0, therefore:

[0164]

[0165]

[0166]

[0167] Based on the above formula and the obtained You can get Numerical solutions for A1 and A1.

[0168] Based on the third constraint, the fitting function End of the first segment The fitting error is ε* ,Right now

[0169]

[0170]

[0171] Based on the above formula, we can obtain Numerical solution. Optional, A1 and Numerical solutions can be obtained through mathematical numerical methods, such as using the vpasolve function in MATLAB software.

[0172] The above fitting function Its equivalent fitting expression is: and,

[0173] Therefore, we can obtain The fitting coefficients are A1 and B1. The process for determining the endpoints and fitting coefficients of other segmented intervals is the same as that for the first segmented interval, and will not be repeated here.

[0174] It should be noted that the fitting error function mentioned above is a relative fitting error function. Alternatively, an absolute fitting error function can also be used, i.e.:

[0175] Using the method described above, given an arbitrary upper limit value ε for the maximum fitting error... * In this case, the starting and ending points of multiple segment intervals in a non-uniformly segmented system, as well as the fitting coefficients for each segment interval, can be determined segment by segment. However, it is possible that the calculated endpoint of the last segment exceeds the range of values ​​for the independent variable, i.e. In this case, the actual maximum fitting error of the last segment is less than the upper limit of the maximum fitting error, i.e. The last segment exhibits overfitting, while the fitting performance of the other segments barely matches the upper limit of the maximum fitting error. Therefore, the actual maximum fitting error of the last segment can be appropriately increased. Consequently, the segment interval of the last segment will increase, while the segment intervals of the other segments will decrease accordingly. This will lower the corresponding maximum fitting error, thus reducing the total maximum fitting error across all segments and achieving a lower upper limit for the maximum fitting error.

[0176] Therefore, based on the maximum fitting error update function g(ε0), the new maximum fitting error ε is determined based on the initial maximum fitting error. * Based on this new maximum fitting error, the independent variable interval is non-uniformly divided. In this embodiment, the new maximum fitting error ε... * The method for determining the new maximum fitting error ε is not specifically limited, as long as it makes the new maximum fitting error ε achievable.* It should be less than the initial maximum fitting error.

[0177] Iterative update ε * Then, the non-uniform segmentation process is repeated until the maximum fitting error of each segment is equal to the updated ε. * That is, it completes the task given ε. * The optimal non-uniform segmentation is found.

[0178] Furthermore, common nonlinear functions often have zeros, making it impossible to solve for the fitting coefficients or resulting in inaccurate values. Therefore, the above process is optimized for the presence of zeros. The optimization principle is to make the zeros part of the fitting function. The intersection point with the target nonlinear function f(x) has zero absolute and relative errors at the zero point, so the relative error can still be used for non-uniform segmentation of the function.

[0179] Specifically, in one case, when the zero point x0 of the function is the starting or ending point of the function to be fitted, i.e., x0 = a or x0 = b, then the above-mentioned first constraint condition is replaced with the fitting function. At the starting point The fitting error is equal to 0, that is:

[0180]

[0181]

[0182]

[0183] Therefore, we get All other conditions remain unchanged.

[0184] like Figure 9 As shown, it illustrates a schematic diagram of a non-uniform segmentation result provided by an embodiment of this application. It can be seen that the non-uniform segmentation processing of nonlinear functions provided by the embodiment of this application can obtain better segmentation results.

[0185] like Figure 10 The diagram illustrates a segmentation result for an interval with zeros as endpoints, provided by an embodiment of this application. It can be seen that when a zero exists and the zero is an endpoint of the range of values ​​for the independent variable, the above-mentioned first constraint condition is replaced by a fitting function. At the starting point By setting the fitting error to zero, the influence of zero points on the segmentation results can be avoided, thus improving the segmentation accuracy.

[0186] In another case, when the zero point x0 of the function is neither the starting point nor the ending point of the target nonlinear function f(x), x0 ≠ a and x0 ≠ b. The fitting coefficient A is determined by the starting point and zero point of the segmented interval. n That is, replace the second constraint mentioned above with the fitted function. The fitting coefficient A in the nth interval where x0 is located n satisfy:

[0187]

[0188] Wherein, the fitting function of the (n-1)th segment Points on Given this, the fitting slope A of the nth interval can be directly obtained. n Other conditions remain unchanged. For example... Figure 11 The diagram illustrates a segmentation result for a non-endpoint interval provided by an embodiment of this application. It is evident that optimal non-uniform segmentation can be achieved in both cases.

[0189] It should be noted that the calculation equations listed above for non-uniform segmentation are only possible examples, and there are many other variations of the specific solution expression.

[0190] Taking the processing of communication signals by a terminal involving power functions as an example, for instance, using... Taking the fitting of x belonging to [0.5,1) as an example, as... Figure 12 As shown, it illustrates a schematic diagram of a uniform segmentation result provided in an embodiment of this application, as follows. Figure 13 The diagram illustrates a segmentation result of a non-uniform segmentation method provided in this embodiment. It is evident that, with equal maximum fitting errors (both 0.1%), the number of segment intervals in the prior art for uniform segmentation is 4, while the number of segments in the non-uniform segmentation method of this embodiment is 3. Consequently, the fitting coefficient obtained by the non-uniform segmentation in this embodiment is relatively small.

[0191] As can be seen, when using the same maximum fitting error upper limit, the non-uniform segmentation method in this embodiment can achieve fewer segments, which helps to reduce the storage overhead of coefficients and the number of operations to determine the segment position. Similarly, when using the same number of segments, the non-uniform segmentation method in this embodiment can achieve a smaller maximum fitting error, which helps to improve the performance of function fitting. Consequently, it makes the chip in the terminal respond to business processing instructions more efficiently and the processing results of communication signals more accurate.

[0192] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0193] Based on the same inventive concept, this application also provides a business data processing apparatus for implementing the business data processing method described above. The solution provided by this apparatus is similar to the implementation scheme described in the above method; therefore, the specific limitations in one or more business data processing apparatus embodiments provided below can be found in the limitations of the business data processing method described above, and will not be repeated here.

[0194] In one embodiment, such as Figure 14 As shown, a business data processing apparatus 1400 is provided, comprising: a first determining module 1401, a second determining module 1402, and a third determining module 1403, wherein:

[0195] The first determining module 1401 is used to receive a service processing instruction and determine the target nonlinear function and the value of the independent variable of the target nonlinear function according to the service processing instruction. The service processing instruction is used to instruct the processing of the communication signal.

[0196] The second determining module 1402 is used to determine the target independent variable interval where the value of the independent variable is located from multiple candidate independent variable intervals corresponding to the target nonlinear function, wherein the multiple candidate independent variable intervals are obtained by non-uniformly dividing the range of independent variable values ​​of the target nonlinear function, and the maximum fitting error corresponding to each candidate independent variable interval is equal.

[0197] The third determining module 1403 is used to determine the target linear fitting coefficients corresponding to the target independent variable interval, and to process the communication signal in response to the service processing instruction based on the target linear fitting coefficients.

[0198] In one embodiment, such as Figure 15As shown, the business data processing device 1500 further includes a processing module 1404, which is used to: perform multiple non-uniform division processing on the range of values ​​of the independent variable according to the initial maximum fitting error and the target nonlinear function, until the iteration stopping condition is met; take the multiple independent variable intervals obtained by non-uniform division processing on the range of values ​​of the independent variable at the time of stopping iteration as the multiple candidate independent variable intervals, and take the linear fitting coefficients corresponding to each independent variable interval at the time of stopping iteration as the linear fitting coefficients corresponding to each candidate independent variable interval.

[0199] In one embodiment, the processing module 1404 is specifically used to: determine whether a characteristic independent variable interval exists among multiple independent variable intervals, the characteristic independent variable interval containing the zero point of the target nonlinear function; if a characteristic independent variable interval exists, adjust the linear fitting coefficients corresponding to the characteristic independent variable interval, and use the adjusted linear fitting coefficients as the linear fitting coefficients of the corresponding candidate independent variable intervals.

[0200] In one embodiment, the processing module 1404 is specifically used to adjust the linear fitting coefficients corresponding to the feature variable intervals, including: recalculating the linear fitting coefficients corresponding to the feature variable intervals based on target constraints when the endpoints of the first or second intervals of the feature variable intervals are zero; the target constraints include that the fitting error of the endpoints of the first intervals of the feature variable intervals is equal to 0.

[0201] In one embodiment, the processing module 1404 is specifically used to adjust the linear fitting coefficients corresponding to the feature variable interval, including: when the feature variable interval contains a zero point and the zero point is not an interval endpoint of the feature variable interval, recalculating the linear fitting coefficients of the feature variable interval based on the zero point, the first interval endpoint of the feature variable interval, and the value of the target nonlinear fitting function at the zero point.

[0202] In one embodiment, the range of values ​​for the independent variable includes the interval between the first endpoint and the second endpoint, and the iteration stopping condition is that the first endpoint is the endpoint of the first independent variable interval among the plurality of independent variable intervals, and the second endpoint is the endpoint of the last independent variable interval among the plurality of independent variable intervals.

[0203] In one embodiment, the processing module 1404 is specifically configured to: determine the target maximum fitting error corresponding to the i-th non-uniform partitioning process based on the initial maximum fitting error, wherein, when i=1, the target maximum fitting error corresponding to the i-th non-uniform partitioning process is the initial maximum fitting error; when i>1, the target maximum fitting error corresponding to the i-th non-uniform partitioning process is determined based on the target maximum fitting error corresponding to the (i-1)-th non-uniform partitioning process; and perform non-uniform partitioning processing on the range of values ​​of the independent variable based on the target maximum fitting error corresponding to the i-th non-uniform partitioning process.

[0204] In one embodiment, the processing module 1404 is specifically configured to: subtract the target maximum fitting error corresponding to the (i-1)th non-uniform partitioning process from the preset step size to obtain the target maximum fitting error corresponding to the i-th non-uniform partitioning process; or, obtain the interval maximum fitting error corresponding to the last independent variable interval, and take the average of the interval maximum fitting error and the target maximum fitting error corresponding to the (i-1)th non-uniform partitioning process as the target maximum fitting error corresponding to the i-th non-uniform partitioning process; or, take the weighted average of the maximum fitting errors corresponding to multiple independent variable intervals obtained by the (i-1)th non-uniform partitioning process as the target maximum fitting error corresponding to the i-th non-uniform partitioning process.

[0205] In one embodiment, the processing module 1404 is specifically configured to: perform multiple interval determination operations on the range of values ​​of the independent variable based on the target maximum fitting error corresponding to the i-th non-uniform partitioning process, to obtain multiple independent variable intervals corresponding to the i-th non-uniform partitioning process, each independent variable interval including the interval range between the first interval endpoint and the second interval endpoint; wherein, the j-th interval determination operation in the multiple interval determination operations includes: when j=1, taking the first endpoint as the first interval endpoint of the j-th independent variable interval; when j>1, taking the second interval endpoint of the (j-1)-th independent variable interval as the first interval endpoint of the j-th independent variable interval; and determining the second interval endpoint of the j-th independent variable interval and the corresponding linear fitting coefficient based on the target maximum fitting error corresponding to the i-th non-uniform partitioning process.

[0206] In one embodiment, the processing module 1404 is specifically configured to: construct a system of equations based on a pre-set first constraint, a second constraint, and a third constraint, and solve the system of equations to obtain the second interval endpoints of the j-th independent variable interval and the corresponding linear fitting coefficients; wherein the first constraint includes that the fitting error of the first interval endpoints of the j-th independent variable interval is equal to the target maximum fitting error; the second constraint includes that the fitting error of one and only one point other than the interval endpoints in the j-th independent variable interval is equal to the target maximum fitting error; and the third constraint includes that the fitting error of the second interval endpoints of the j-th independent variable interval is equal to the target maximum fitting error.

[0207] In one embodiment, the processing module 1404 is specifically configured to: obtain a candidate maximum fitting error, determine whether the candidate maximum fitting error meets a preset condition; if the preset condition is met, use the candidate maximum fitting error as the initial maximum fitting error; if the preset condition is not met, adjust the candidate maximum fitting error until the adjusted candidate maximum fitting error meets the preset condition; the preset condition includes: the number of multiple reference independent variable intervals obtained by non-uniformly dividing the range of values ​​of the independent variable is equal to a preset target number, and the target number is an integer multiple of the preset value.

[0208] In one embodiment, the processing module 1404 is specifically used to: increase the candidate maximum fitting error if the number of multiple reference independent variable intervals obtained by non-uniformly dividing the range of values ​​of the independent variable is greater than the target number; and decrease the candidate maximum fitting error if the number of multiple reference independent variable intervals obtained by non-uniformly dividing the range of values ​​of the independent variable is less than the target number.

[0209] In one embodiment, the third determining module 1403 is specifically used to: determine a linear fitting function based on the target linear fitting coefficients; and process the communication signal in response to the service processing instruction using the linear fitting function.

[0210] Each module in the aforementioned business data processing device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the operations corresponding to each module.

[0211] In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 16As shown, the computer device includes a processor, memory, and a network interface connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The database stores business data processing data. The network interface communicates with external terminals via a network connection. When the computer program is executed by the processor, it implements a business data processing method.

[0212] In one embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 17 As shown, the computer device includes a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, NFC (Near Field Communication), or other technologies. When the computer program is executed by the processor, it implements a business data processing method. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad mounted on the computer device casing, or an external keyboard, touchpad, or mouse.

[0213] Those skilled in the art will understand that Figure 16 and 17 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0214] In one embodiment, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above method embodiments.

[0215] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon that, when executed by a processor, implements the steps in the above method embodiments.

[0216] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0217] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0218] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0219] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A business data processing method, characterized in that, The method includes: Receive a service processing instruction, determine the target nonlinear function and the value of the independent variable of the target nonlinear function according to the service processing instruction, wherein the service processing instruction is used to instruct the processing of the communication signal; From the multiple candidate independent variable intervals corresponding to the target nonlinear function, the target independent variable interval in which the value of the independent variable is located is determined. The multiple candidate independent variable intervals are obtained by non-uniformly dividing the range of the independent variable values ​​of the target nonlinear function, and the maximum fitting error corresponding to each candidate independent variable interval is equal. Determine the target linear fitting coefficients corresponding to the target independent variable interval, and process the communication signal in response to the service processing instruction based on the target linear fitting coefficients; The method further includes: performing multiple non-uniform partitioning processes on the range of values ​​of the independent variable based on the initial maximum fitting error and the target nonlinear function, until the iteration stopping condition is met, wherein the range of values ​​of the independent variable includes the interval between a first endpoint and a second endpoint, and the iteration stopping condition is that the first endpoint is the endpoint of the first independent variable interval among the plurality of independent variable intervals, and the second endpoint is the endpoint of the last independent variable interval among the plurality of independent variable intervals; taking the plurality of independent variable intervals obtained by performing non-uniform partitioning processes on the range of values ​​of the independent variable at the time of stopping iteration as the plurality of candidate independent variable intervals, and taking the linear fitting coefficients corresponding to each of the independent variable intervals at the time of stopping iteration as the linear fitting coefficients corresponding to each of the candidate independent variable intervals; The iterative execution of the i-th non-uniform partitioning process in multiple non-uniform partitioning processes includes: determining the target maximum fitting error corresponding to the i-th non-uniform partitioning process based on the initial maximum fitting error, wherein, when i=1, the target maximum fitting error corresponding to the i-th non-uniform partitioning process is the initial maximum fitting error; when i>1, the target maximum fitting error corresponding to the i-th non-uniform partitioning process is determined based on the target maximum fitting error corresponding to the (i-1)-th non-uniform partitioning process; and performing non-uniform partitioning on the range of values ​​of the independent variable based on the target maximum fitting error corresponding to the i-th non-uniform partitioning process.

2. The method according to claim 1, characterized in that, The step of using the linear fitting coefficients corresponding to each of the independent variable intervals at the time of stopping iteration as the linear fitting coefficients corresponding to each of the candidate independent variable intervals includes: Determine whether a characteristic independent variable interval exists among the plurality of independent variable intervals, wherein the characteristic independent variable interval contains the zeros of the target nonlinear function; If the specified feature variable interval exists, the linear fitting coefficients corresponding to the specified feature variable interval are adjusted, and the adjusted linear fitting coefficients are used as the linear fitting coefficients of the corresponding candidate variable intervals.

3. The method according to claim 2, characterized in that, The adjustment process for the linear fitting coefficients corresponding to the intervals of the characteristic independent variables includes: If the endpoint of the first interval or the endpoint of the second interval of the characteristic independent variable interval is the zero point, the linear fitting coefficients corresponding to the characteristic independent variable interval are recalculated based on the target constraint condition; the target constraint condition includes that the fitting error of the first interval endpoint of the characteristic independent variable interval is equal to 0.

4. The method according to claim 2, characterized in that, The adjustment process for the linear fitting coefficients corresponding to the intervals of the characteristic independent variables includes: If the zero point is included in the interval of the characteristic independent variable and the zero point is not an interval endpoint of the interval of the characteristic independent variable, the linear fitting coefficient of the interval of the characteristic independent variable is recalculated based on the zero point, the first interval endpoint of the interval of the characteristic independent variable, and the value of the target nonlinear fitting function at the zero point.

5. The method according to claim 1, characterized in that, When i is greater than 1, the process of determining the maximum fitting error of the target corresponding to the i-th non-uniform partitioning process includes: The maximum fitting error corresponding to the (i-1)th non-uniform partitioning process is obtained by subtracting the preset step size from the maximum fitting error corresponding to the i-th non-uniform partitioning process. Alternatively, obtain the maximum fitting error of the interval corresponding to the last independent variable interval, and take the average of the maximum fitting error of the interval and the target maximum fitting error corresponding to the (i-1)th non-uniform partitioning process as the target maximum fitting error corresponding to the i-th non-uniform partitioning process. Alternatively, the weighted average of the maximum fitting errors corresponding to the multiple independent variable intervals obtained from the (i-1)th non-uniform partitioning process can be used as the target maximum fitting error corresponding to the i-th non-uniform partitioning process.

6. The method according to claim 1, characterized in that, The step of performing non-uniform partitioning of the independent variable's value range based on the target maximum fitting error corresponding to the i-th non-uniform partitioning process includes: Based on the target maximum fitting error corresponding to the i-th non-uniform partitioning process, multiple interval determination operations are performed on the range of values ​​of the independent variable to obtain multiple independent variable intervals corresponding to the i-th non-uniform partitioning process. Each independent variable interval includes the interval range between the endpoints of the first interval and the second interval. Among them, the j-th interval determination operation in the multiple interval determination operations includes: When j=1, the first endpoint is taken as the first interval endpoint of the j-th independent variable interval; when j>1, the second interval endpoint of the (j-1)-th independent variable interval is taken as the first interval endpoint of the j-th independent variable interval. The endpoints of the second interval of the j-th independent variable interval and the corresponding linear fitting coefficients are determined based on the target maximum fitting error corresponding to the i-th non-uniform partitioning process.

7. The method according to claim 6, characterized in that, The step of determining the second interval endpoints and corresponding linear fitting coefficients of the j-th independent variable interval based on the target maximum fitting error corresponding to the i-th non-uniform partitioning process includes: A system of equations is constructed based on the pre-defined first, second, and third constraints, and the system of equations is solved to obtain the second interval endpoints of the j-th independent variable interval and the corresponding linear fitting coefficients. The first constraint condition includes that the fitting error of the first interval endpoint of the j-th independent variable interval is equal to the target maximum fitting error; the second constraint condition includes that the fitting error of one and only one point other than the interval endpoint in the j-th independent variable interval is equal to the target maximum fitting error; and the third constraint condition includes that the fitting error of the second interval endpoint of the j-th independent variable interval is equal to the target maximum fitting error.

8. The method according to claim 1, characterized in that, The process of determining the initial maximum fitting error includes: Obtain candidate maximum fitting errors and determine whether the candidate maximum fitting errors meet preset conditions. If the preset conditions are met, the candidate maximum fitting error is taken as the initial maximum fitting error; If the preset conditions are not met, the candidate maximum fitting error is adjusted until the adjusted candidate maximum fitting error meets the preset conditions. The preset conditions include: the number of multiple reference independent variable intervals obtained by non-uniformly dividing the range of values ​​of the independent variable is equal to the preset target number, and the target number is an integer multiple of the preset value.

9. The method according to claim 8, characterized in that, The adjustment process for the candidate maximum fitting error includes: If the number of reference independent variable intervals obtained by non-uniformly dividing the range of values ​​of the independent variable is greater than the target number, then the candidate maximum fitting error is increased. If the number of reference independent variable intervals obtained by non-uniformly dividing the range of values ​​of the independent variable is less than the target number, then the candidate maximum fitting error is reduced.

10. The method according to claim 1, characterized in that, The step of processing the communication signal in response to the service processing instruction based on the target linear fitting coefficients includes: Determine the linear fitting function based on the target linear fitting coefficients; The communication signals are processed in response to the service processing instructions using the linear fitting function.

11. A business data processing apparatus, characterized in that, The device includes: The first determining module is used to receive a service processing instruction and determine the target nonlinear function and the value of the independent variable of the target nonlinear function according to the service processing instruction. The service processing instruction is used to instruct the processing of the communication signal. The second determining module is used to determine the target independent variable interval where the value of the independent variable is located from multiple candidate independent variable intervals corresponding to the target nonlinear function, wherein the multiple candidate independent variable intervals are obtained by non-uniformly dividing the range of independent variable values ​​of the target nonlinear function, and the maximum fitting error corresponding to each candidate independent variable interval is equal. The third determining module is used to determine the target linear fitting coefficients corresponding to the target independent variable interval, and to process the communication signal in response to the service processing instruction based on the target linear fitting coefficients; The device further includes a processing module; the processing module is configured to: iteratively perform multiple non-uniform partitioning processes on the range of values ​​of the independent variables based on the initial maximum fitting error and the target nonlinear function, until the iteration stopping condition is met, wherein the range of values ​​of the independent variables includes the interval between a first endpoint and a second endpoint, and the iteration stopping condition is that the first endpoint is the endpoint of the first independent variable interval among the plurality of independent variable intervals, and the second endpoint is the endpoint of the last independent variable interval among the plurality of independent variable intervals; and use the plurality of independent variable intervals obtained by non-uniform partitioning the range of values ​​of the independent variables at the time of stopping iteration as the plurality of candidate independent variable intervals, and set the values ​​of each independent variable interval at the time of stopping iteration as the candidate independent variable intervals. The corresponding linear fitting coefficients are used as the linear fitting coefficients for each candidate independent variable interval; wherein, the i-th non-uniform partitioning process in the iterative execution of multiple non-uniform partitioning processes includes: determining the target maximum fitting error corresponding to the i-th non-uniform partitioning process based on the initial maximum fitting error, wherein, when i=1, the target maximum fitting error corresponding to the i-th non-uniform partitioning process is the initial maximum fitting error; when i>1, the target maximum fitting error corresponding to the i-th non-uniform partitioning process is determined based on the target maximum fitting error corresponding to the (i-1)-th non-uniform partitioning process; and performing non-uniform partitioning processing on the range of values ​​of the independent variable based on the target maximum fitting error corresponding to the i-th non-uniform partitioning process.

12. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 10.

13. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 10.

14. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 10.

Images