Vehicle power optimization fuzzy control system and method based on optimal energy management, storage medium and computer program product

Abstract

The invention provides a vehicle power optimization fuzzy control system and method based on optimal energy management, a storage medium and a computer program product. The system comprises an energy management optimization model construction module, an optimal energy distribution database generation module, a fuzzy set and membership function division module, a fuzzy rule generation module and a control output module. By means of the method, accurate optimal distribution of the power of the vehicle power system can be achieved, fuel consumption and power consumption can be effectively reduced, and the energy utilization efficiency is improved; and meanwhile, the stability and adaptability of power output are enhanced, and it is ensured that the vehicle can operate efficiently under different working conditions.

Description

Technical Field

[0001] This invention belongs to the field of automotive powertrain energy management technology, specifically relating to a fuzzy control system, method, storage medium, and computer program product for vehicle power optimization based on optimal energy management. Background Technology

[0002] In recent years, the optimization control of energy management in the power system of new energy vehicles has become a research hotspot for improving vehicle economy and range. Currently, there are two main approaches: one is global optimal algorithms such as dynamic programming (DP), which can achieve theoretically optimal energy allocation, but lacks real-time performance and online adaptability; the other is knowledge- and experience-based rule-based methods such as fuzzy control, which have good real-time performance, but strategy design heavily relies on expert manual setting, making it difficult to guarantee tracking the theoretical optimum, and exhibiting weak adaptability to changing operating conditions. Existing fuzzy controllers generally suffer from the following limitations: membership functions and rule base divisions are coarse, mainly determined by subjective judgment based on experience; the rule structure is rigid and has poor portability, lacking intelligent data-driven automatic strategy generation. This makes it difficult for traditional fuzzy control to achieve ideal results in complex, high-dimensional, and multi-objective energy management scenarios. Summary of the Invention

[0003] To address the problems in vehicle energy management where fuzzy controllers rely on manual settings, are unsuitable for complex operating conditions, and struggle to adequately approximate theoretical optimality, this invention proposes a fuzzy control system, method, storage medium, and computer program product for vehicle power optimization based on optimal energy management.

[0004] A vehicle power optimization fuzzy control system based on optimal energy management, which achieves one of the objectives of this invention, includes: Energy Management Optimization Model Construction Module: Used to establish an energy management optimization model that includes power demand, battery status, and energy source constraint parameters based on vehicle operating condition data and vehicle power system model; The optimal energy allocation database generation module is used to divide the energy management process of the vehicle operating cycle into several sub-processes according to the time step based on the objective function minimization criterion of the energy management optimization model and the global optimization method. Following the power demand, battery state and energy source constraint parameters in the energy management optimization model, it enumerates in reverse order to calculate the optimal fuel consumption and electricity consumption schemes for each sub-process, generating a theoretically optimal energy allocation sequence. The global optimization method is a combination of numerical recursion and trajectory recovery based on the Bellman optimality principle. The energy source constraint parameters include the upper and lower limits of battery SOC, the maximum output power of the engine, and the maximum charging and discharging power of the battery. The fuzzy set and membership function partitioning module is used to perform unsupervised classification of the input and output variables in the theoretically optimal energy allocation sequence using an improved K-means algorithm, obtaining fuzzy sets for each input variable and each output variable, including their value range, mean, and standard deviation. These fuzzy sets are predefined classification sets of the variable space, obtained by clustering variables in the optimal energy allocation database using the improved K-means algorithm. These discrete fuzzy categories, such as low (L), medium (M), and high (H), serve as the basic linguistic terms for fuzzy rules, used to construct the mapping relationship between inputs and outputs. Membership functions are constructed based on the mean and standard deviation. The membership functions determine the fuzzy set to which each input and output variable in the theoretically optimal energy allocation sequence belongs. The input variables include demand power. P DR The power battery's state of charge (SOC) and the output variables include engine power. P eng or fuel cell power P fc The membership function can be a Gaussian membership function or other membership functions; this invention does not limit this. The fuzzy rule generation module is used to count the frequency of the combination of input variable fuzzy sets corresponding to each output variable fuzzy set; taking each fuzzy set classification of the output variable as the root node of the decision tree, selecting two input variable fuzzy sets from the input variables as the second-level and third-level nodes of the decision tree respectively, and filtering out the target second-level node and target third-level node from the second-level and third-level nodes based on the frequency; extracting the mapping relationship between the input variable fuzzy sets and the output variable fuzzy sets from the root node, target second-level node, and target third-level node, traversing each fuzzy set classification of the output variable and the combination of input variables, and finally obtaining the fuzzy rule set composed of the mapping relationship; Control output module: Used to output energy distribution control commands for the vehicle power system according to the fuzzy rule set, regulate the output power of the engine / fuel cell and power battery, and ensure that the total output power of the energy source meets the sum of the vehicle's required power and transmission loss power.

[0005] Furthermore, the objective function expression of the energy management optimization model includes: In the formula, J represents the total energy consumption cost of the entire operating cycle; T represents the total duration of the operating cycle. P eng ( t () represents the output power of the engine / fuel cell at time t; f ( P eng (t ))express t Fuel consumption rate of the engine / fuel cell at any given time; P bat ( t () represents the charging / discharging power of the power battery at time t; eP bat ( t )) represents the power battery's energy consumption rate at time t; w 1 and w 2 represents the weighting coefficients used to unify fuel and electricity consumption dimensions, calibrated based on actual vehicle models.

[0006] Furthermore, the vehicle powertrain model includes an energy source output characteristics and constraint sub-model, a vehicle power demand calculation sub-model, and a transmission system power loss sub-model. The energy source output characteristics and constraint sub-model is used to provide the energy consumption calculation basis and safety constraint boundary for the energy management optimization model. The vehicle power demand calculation sub-model is used to provide the overall power allocation target for the energy management optimization model, and the transmission system power loss sub-model is used to provide power balance constraints for the energy management optimization model.

[0007] Furthermore, in the fuzzy rule generation module, the method for sequentially filtering target second-level nodes and target third-level nodes from second-level nodes and third-level nodes based on the frequency includes: Select an input variable from the input variables, and use the fuzzy set of the input variable as a candidate for the second-level node of the decision tree. Based on the frequency, select the fuzzy set category with the highest combination frequency with the root node from the candidates for the second-level node as the target second-level node. Select one input variable from the remaining input variables, and use the fuzzy set of this input variable as a candidate for a third-level node in the decision tree. Based on the frequency, select the fuzzy set category with the highest combination frequency with the root node from the candidates for third-level nodes as the target third-level node. If there are fuzzy set categories with the same combination frequency, select the fuzzy set category with the smallest sample variance in the corresponding optimal energy allocation database as the target node.

[0008] Furthermore, the method for traversing each fuzzy set classification of the output variable and the combination of input variables in the fuzzy rule generation module includes: Iterate through each fuzzy set classification of the output variable and the different combinations of all input variables in order; the different combinations of all input variables in order include: all possible permutations of each input variable as a second-level node and other input variables as third-level nodes, ensuring that all combinations of input variables at the second-level and third-level node positions are covered.

[0009] Furthermore, the method for unsupervised classification of the input and output variables in the theoretically optimal energy allocation sequence includes: S1. Randomly select k samples from the dataset of the corresponding variable in the optimal energy allocation database as cluster centers; S2. Calculate the Euclidean distance between each sample in the dataset and all cluster centers, and assign each sample to the cluster containing the nearest cluster center. The range of sample values ​​in each cluster is the initial range of the fuzzy set. S3. Calculate the arithmetic mean of all samples within each cluster to obtain the updated cluster center; S4. If the difference between the updated cluster centers and the previous cluster centers is less than or equal to the set value, then end the current clustering process; otherwise, proceed to step S3. S5. Select the result with the smallest sum of intra-cluster distances from the clustering results as the fuzzy set of each input variable and the fuzzy set of each output variable, including the range of values, mean, and standard deviation.

[0010] Furthermore, methods for determining the number of clusters k include: Calculate the sum of Euclidean distances between all samples within a cluster and the cluster center for different k values, plot the curve of the sum of distances as a function of k, and take the k value when the slope of the sum of distances between adjacent k values ​​is less than a set proportion as the number of clusters.

[0011] A second objective of this invention is a fuzzy control method for vehicle power optimization based on optimal energy management, comprising: Based on vehicle operating condition data and vehicle power system model, an energy management optimization model is established that includes power demand, battery status, and energy source constraint parameters. Based on the objective function minimization criterion of the energy management optimization model, the energy management process of the vehicle operating cycle is divided into several sub-processes according to the time step using a global optimization method. Following the power demand, battery status and energy source constraint parameters in the energy management optimization model, the optimal fuel consumption and electricity consumption schemes of each sub-process are calculated in reverse order to generate the theoretically optimal energy allocation sequence. An improved K-means algorithm is used to perform unsupervised classification of the input and output variables in the theoretically optimal energy allocation sequence, resulting in fuzzy sets for each input variable and each output variable, including their value range, mean, and standard deviation. A membership function is constructed based on the mean and standard deviation. The fuzzy set to which each input and output variable in the theoretically optimal energy allocation sequence belongs is determined based on the membership function. The frequency of input variable fuzzy set combinations corresponding to each output variable fuzzy set is counted. Taking each fuzzy set classification of the output variable as the root node of the decision tree, two input variable fuzzy sets are selected from the input variables as the second-level and third-level nodes of the decision tree, respectively. Based on the frequency, the target second-level node and target third-level node are sequentially selected from the second-level and third-level nodes. The mapping relationship between the input variable fuzzy sets and the output variable fuzzy sets is extracted from the root node, the target second-level node, and the target third-level node. Each fuzzy set classification of the output variable and the combination of input variables are traversed to finally obtain the fuzzy rule set composed of the mapping relationship. The energy distribution control command of the vehicle power system is output according to the fuzzy rule set to regulate the output power of the engine / fuel cell and the power battery.

[0012] A non-transitory computer-readable storage medium for achieving the third objective of the present invention stores a computer program thereon, characterized in that, when the computer program is executed by a processor, it implements the steps of the vehicle power optimization fuzzy control method based on optimal energy management.

[0013] A computer program product for achieving the fourth objective of the present invention includes a computer program / instruction that, when executed by a processor, implements the steps of the vehicle power optimization fuzzy control method based on optimal energy management.

[0014] The beneficial effects of this invention include: 1. By improving the K-means algorithm, the input / output variable fuzzy sets are automatically divided, and the fuzzy rules are summarized based on the optimal data using the decision tree algorithm. This eliminates the need for manually setting membership functions and rules, thus solving the subjectivity defects of traditional fuzzy control. 2. A global optimization method is used to generate the theoretically optimal energy distribution data sequence for all operating conditions. Based on this, the optimal data generation fuzzy rule is obtained, which makes the energy distribution adapt to the dynamic operating conditions of the vehicle, reduces the total fuel and electricity consumption, and improves the driving range. 3. Fuzzy sets can adaptively adjust according to vehicle parameters and operating condition data, adapting to changing operating conditions without repeated development, thus improving the flexibility of the solution; 4. Built-in constraints on battery SOC, engine / fuel cell power, etc., prevent overcharging, over-discharging, overload or inefficient operation, while ensuring that energy output covers demand and losses, and protects critical components. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of the system described in this invention; Figure 2 This is a schematic diagram of the decision-making power described in this invention. Detailed Implementation

[0016] The following detailed embodiments are provided to explain the technical solutions of the present invention, so that those skilled in the art can understand the present invention. The scope of protection of the present invention is not limited to the following specific embodiments. Any modifications or improvements made by those skilled in the art that incorporate the technical solutions of the present invention but differ from the following detailed embodiments are also within the scope of protection of the present invention.

[0017] A fuzzy control method for vehicle power optimization based on optimal energy management includes the following steps: 1. Calculate the optimal energy allocation database. First, using known vehicle powertrain models and operating condition data, an energy management optimization model is established, including parameters such as power demand, battery status, and energy source constraints.

[0018] The energy management optimization model is used to calculate the optimal energy allocation scheme for energy sources such as engines / fuel cells, based on the criterion of minimizing the objective function, under the premise of meeting vehicle driving requirements and energy source safety constraints. It transforms energy management optimization into a solvable mathematical problem, ultimately achieving the minimum total energy consumption.

[0019] In one embodiment, the input data includes: operating condition data, real-time status parameters, and system constraint parameters; the output data is the optimal fuel consumption and electricity consumption scheme for each subprocess, such as the output power allocation value of each energy source and the fuel consumption corresponding to the allocation scheme; the output power allocation value is, for example, the engine output power P. eng (kW) and battery discharge P bat (kW); the fuel consumption corresponding to the allocation plan is as follows: fuel consumption L eng (g / s) and power consumption L bat (Wh / s).

[0020] In one embodiment, the operating condition data refers to key data that reflects the actual driving status of the vehicle and determines the power demand, including: vehicle speed-time curve, such as the process of the vehicle speed accelerating from 0 to 30km / h and 25km / h and then decelerating back to 0 in a certain operating condition, which directly determines the power demand; it also includes: driving time, gradient change data, ambient temperature, vehicle load; current SOC of the power battery, and current efficiency of the engine / fuel cell.

[0021] In one embodiment, the system constraint parameters include: upper and lower limits of battery SOC, maximum engine output power, and maximum battery charge and discharge power.

[0022] The vehicle powertrain system model is used to describe the energy output characteristics, state change patterns, and constraint boundaries of each core component of the vehicle powertrain system, and can support a multi-sub-model integrated system for energy management optimization calculations. The construction of the energy management optimization model is based on the sub-models of the vehicle powertrain system model. Each sub-model provides the basis for calculating the target energy consumption of power allocation and the safety constraint boundaries for the energy management optimization model. Together, they constitute the calculation logic chain for optimal energy allocation.

[0023] In one embodiment, the vehicle powertrain model includes an energy source output characteristics and constraint sub-model, a vehicle power demand calculation sub-model, and a transmission system power loss sub-model. The energy source output characteristics and constraint sub-model is used to provide the energy consumption calculation basis and safety constraint boundary for the energy management optimization model. The vehicle power demand calculation sub-model is used to provide the overall power allocation target for the energy management optimization model, and the transmission system power loss sub-model is used to provide power balance constraints for the energy management optimization model.

[0024] In one embodiment, the energy source output characteristics and constraint sub-model are used to define the energy output capability and efficiency characteristics of the engine or fuel cell and power battery, specifically including: Engine / Fuel Cell Power-Efficiency Sub-Model: Used to quantify the relationship between the output power and energy consumption rate of an active energy source; for an engine, its output power... P eng With fuel consumption rate f ( P eng Satisfies the quadratic fitting relationship: Where a, b, and c are constants calibrated based on specific engine bench test data; if it is a fuel cell, its output power... P fcg With hydrogen consumption rate h(P fc ) Satisfying similar fitting relationship d, e, and f are the experimental calibration parameters for the fuel cell. The constraint boundary is: the output power of the active energy source must satisfy... P min ≤P eng / P fc ≤P max ,in P min Minimum output power to avoid engine shutdown / inefficient operation; P maxThe rated maximum output power is determined. The power-efficiency relationship of this sub-model provides the energy consumption calculation basis for the objective function of the energy management optimization model, and the constraint boundary defines the safe range for the optimization process of the energy management optimization model, ensuring that the generated energy allocation scheme will not lead to inefficient engine operation or fuel cell overload.

[0025] In one embodiment, the power battery state and charge / discharge constraint sub-model is used to calculate the real-time changes in the remaining SOC of the power battery, define the safety boundary of the battery's charge / discharge power, and avoid damage to the battery from overcharging and over-discharging. It is a key constraint on "whether the battery is charged / discharged and how much it is charged / discharged" during energy allocation; its model expression includes: SOC(t) is the battery state of charge at time t; SOC(t0) is the initial state of charge, Q bat U is the rated capacity of the battery. bat This is the battery's rated voltage. for The battery charging and discharging power is displayed at all times, with discharging being positive and charging being negative. This is a dynamically adjusted battery charge / discharge efficiency based on ambient temperature; its constraints include: the battery's State of Charge (SOC) must meet the following requirements: SOC min ≤SOC(t)≤SOC max The charging and discharging power must meet the following requirements: P bat,max,ch ≤P bat (t)≤P bat,max,dch ; P bat,max,ch Maximum charging power, P bat,max,dch This represents the maximum discharge power. The SOC variation calculated by this sub-model is the core basis for the energy management optimization model to determine whether the battery is chargeable / dischargeable. The charge / discharge power constraint is the constraint condition of the energy management optimization model, ensuring that the optimized battery power allocation meets safety standards.

[0026] In one embodiment, the vehicle power demand calculation sub-model is used to convert vehicle driving condition data into the real-time total power demand of the vehicle. P DR Its calculation formula includes: F roll For rolling resistance, F roll = m × g × f ; m For the overall vehicle quality, gIt is the acceleration due to gravity. f This is the rolling resistance coefficient; F air For air resistance; F grade For slope resistance, This sub-model is used to adjust for power losses during transmission, improving the efficiency of the transmission system. It calculates the total vehicle power demand. P DR This is the overall power allocation objective of the energy management optimization model. All subsequent power allocation calculations for engines and batteries in the energy management optimization model are to meet this objective. P DR This is a prerequisite for ensuring that the optimization results are consistent with the actual driving needs of the vehicle.

[0027] In one embodiment, the power loss sub-model of the transmission system is used to quantify the power loss during the transfer of power from the energy source to the wheels, serving as a crucial supplement to ensure a balanced power distribution. The actual usable power after the energy source output power is transmitted through the transmission system is calculated to ensure that the energy distribution result satisfies the condition that the total output power of the energy source is greater than or equal to the sum of the vehicle's required power and the transmission loss power. The calculation formula includes: Transmission loss power P total This is the sum of the output power of the engine / fuel cell and the power battery. , must meet .

[0028] The transmission loss power calculated by this sub-model P loss and power balance relationship P total = P DR + P loss This is the core basis for the energy management optimization model to verify whether the output power of the energy source is sufficient. If the energy management optimization model generates... P total Unable to cover P DR and P loss If the sum of the values ​​is less than 1, the allocation scheme will be deemed invalid and needs to be re-optimized.

[0029] Based on the objective function minimization criterion, a global optimization method combining numerical recursion and trajectory recovery is employed. According to the Bellman optimality principle, the entire energy management process of a vehicle within a complete operating cycle is divided into several sub-processes according to time steps. In a MATLAB program, all possible energy allocation paths for each sub-process are calculated in reverse order, such as traversing the engine power range from 0 to 60 kW with a step size of 1 kW. This yields the optimal fuel and electricity consumption schemes for each sub-process. While satisfying system constraints, the optimal schemes for each sub-process are combined to generate the theoretically optimal energy allocation sequence for the entire operating cycle. This sequence is stored as an optimal energy allocation database, providing raw sample data for subsequent data-driven controller automatic generation and learning.

[0030] In one embodiment, the objective function includes: In the formula, J represents the total energy consumption cost of the entire operating cycle; T represents the total duration of the operating cycle. P eng ( t () represents the output power of the engine / fuel cell at time t; f ( P eng ( t ))express t Fuel consumption rate of engine / fuel cell at any time, in g / s, varies with output power; the higher the engine power, the higher the fuel consumption rate. P bat ( t () represents the charging / discharging power of the power battery at time t; eP bat ( t )) represents the power battery's energy consumption rate at time t; w 1 and w 2 represents the weighting coefficient, used to unify the dimensions of fuel and electricity consumption. For example, 1g of fuel is equivalent to about 10Wh of electricity. The coefficient is set according to the actual vehicle model to ensure that the target is focused on minimizing total energy consumption.

[0031] In one embodiment, the global optimization method combining numerical recursion and trajectory recovery includes the following steps: The entire operating cycle T is divided into N sub-processes according to a set time step (e.g., 1 second), starting from the end point T of the operating cycle. N The calculation is performed in reverse order from the starting point T1 of the working condition. For each subprocess, all possible energy distribution paths are traversed, such as all values ​​of engine power from 0 to maximum output. The energy consumption of each path is calculated, and the optimal path of the current subprocess, that is, the path with the minimum energy consumption, is selected. The optimal paths of each subprocess are chained together in order from start to finish to form the theoretically optimal energy allocation trajectory for the entire operating cycle. This is the theoretically optimal energy allocation sequence. An example of the allocation sequence is shown in Table 1 below: Table 1 The optimality of this sequence is reflected in the following: when the power demand is low (t1, t2, t8, t9), the battery is used to avoid inefficient engine operation and reduce fuel consumption; when the power demand is medium to high (t3-t7), the engine is started and the optimal power is matched to balance fuel and electricity consumption, and the SOC is always within the constraints.

[0032] 2. Data-driven automatic partitioning of fuzzy sets and membership functions Based on the optimal energy allocation database, an improved K-means algorithm is used to perform unsupervised classification of input variables (including demand power and battery SOC) and output variables (power allocation of the engine / fuel cell). The variable space is adaptively divided into a suitable number and range of fuzzy sets, the number of which is determined by the number of clusters, k. Each cluster is automatically assigned a Gaussian membership function, and the center and width parameters of the membership function are set according to the mean and variance of each cluster, eliminating the need for subjective manual partitioning. Traditional k-means clustering is sensitive to initial cluster centers, which can lead to instability in the clustering results and affect the clustering effect. To ensure consistency across multiple clustering results, the number of clustering iterations is set to n, and the sum of the distances between the sample vectors and the mean vector within each cluster is calculated after each clustering iteration. Then, the sums of these distances are added together, and the final cluster centers are the k cluster centers that minimize the sum of their distances after n clustering iterations.

[0033] The specific solution steps of the algorithm include: 1) Determine the number of clusters k (corresponding to the number of fuzzy sets); In one embodiment, the method for determining the value of k includes: calculating the sum of Euclidean distances between all samples within a cluster and the cluster center for different cluster numbers k (e.g., k=2,3,...,8); plotting a curve of the distance sum changing with k; and when k increases to a certain value, the decreasing slope of the distance sum decreases sharply, forming an inflection point (e.g., the decreasing slope of the distance sum between adjacent K values ​​is <10%, which is considered an inflection point). The corresponding k at this point is the number of clusters. For example, power demand... P DR In the clustering, an inflection point occurs when k=4, therefore... P DR It is divided into 4 fuzzy sets, such as: SL=0~30kW, S=30~60kW, M=60~90kW, H=90~120kW.

[0034] 2) Divide the range of fuzzy sets by improving the K-means algorithm. In one embodiment, the partitioning method includes: The number of clusters is determined based on a specific k value. In the optimal energy allocation database dataset, k data samples are randomly selected and used directly as the initial cluster centers, i.e., the initial mean vectors, to mark the initial position of each cluster; where the dataset is a subset split by variables, if the current cluster object is the required power... P DR Then the dataset is a set of sampled values ​​from all sub-processes in the database; 3) Calculate the distance between each sample in the dataset and all cluster centers, find the cluster center with the closest Euclidean distance for each sample, and assign the sample to the corresponding cluster; the range of sample values ​​for each cluster is the initial range of the fuzzy set. For example, if the sample value of a certain cluster is 0~30kW, the initial range is 0~30kW. 4) Recalculate the new cluster centers within each cluster; calculate the arithmetic mean of all samples in each cluster, and use this mean as the updated cluster center to replace the previous cluster center, such as... P DR A certain cluster sample has a power of 0~30kW, with an average power of 15kW. After the update, the cluster center is 15kW. 5) Determine whether the updated cluster centers are equal to or less than the previous round's cluster centers, or if the difference is less than a set threshold of 10. -3 If the cluster centers are not equal or the difference is greater than the set threshold, then repeat steps 3-4 with the updated cluster centers until the updated cluster centers are equal to or the difference is less than the set threshold. -3 This concludes the clustering process, and the sample data for each cluster are recorded, along with the calculated distances between clusters.

[0035] 6) Repeat steps 2-5 until the set number of clustering iterations n is completed. Select the clustering result corresponding to the minimum inter-cluster distance as the final partition. At this point, the range of sample values ​​for each cluster (i.e., the range between the minimum and maximum values ​​of samples within the cluster) is the range of the fuzzy set. For example, P = the final fuzzy set is SL (0~30kW), S (30~60kW), M (60~90kW), H (90~120kW); SOC's final fuzzy set is L (0.2~0.4), M (0.4~0.6), H (0.6~0.8). P eng The final fuzzy set is L (0~20kW), M (20~40kW), H (40~60kW).

[0036] With input variable demand power P DR For example, see the following example: Sample data for a complete operating condition was extracted from the optimal energy allocation database. The data range is 0~120kW, with a total of 1180 sample values, corresponding to 1180 subprocesses.

[0037] The sum of Euclidean distances between all samples within a cluster and the cluster center for different k values ​​is calculated, and the results are as follows: When K=2, the sum of intra-cluster distances S=1280; when K=3, the sum of intra-cluster distances S=860; when K=4, the sum of intra-cluster distances S=520; when K=5, the sum of intra-cluster distances S=505; when K=6, the sum of intra-cluster distances S=498. It can be seen that when k increases from 3 to 4, S decreases from 860 to 520, a decrease of 40.7%, while when k increases from 4 to 5, S decreases by 2.9%. Therefore, K=4 is determined, which represents the required power. P DR Corresponding to 4 fuzzy sets; From demand power P DR Four initial cluster centers were randomly selected from the data: μ1=20kW, μ2=50kW, μ3=80kW, and μ4=110kW; the power of each cluster center was calculated. P DR The Euclidean distance between the data sample and the four cluster centers mentioned above, as shown in the sample... P DR =25kW is closest to μ1, so it is assigned to the cluster containing μ1; Next, the cluster centers are updated based on the divided clusters. The cluster center is calculated as the mean of the samples within the cluster, such as: Cluster 1: Sample range 0~30kW, mean updated to μ1 ’ =15kW; Cluster 2: Sample range 30~60kW, mean updated to μ2 ’ =45kW; Cluster 3: Sample range 60~90kW, mean updated to μ3 ’ =75kW; Cluster 4: Sample range 90~120kW, mean updated to μ4 ’ =105kW; Calculate the difference between the updated center and the initial center. If it is greater than the set threshold, it means that convergence is not met, and the iteration is repeated.

[0038] After performing multiple clustering operations, the result with the minimum sum of intra-cluster distances is selected as the final clustering, as shown in Table 2 below: Table 2 at this time, P DR The ranges of the four fuzzy sets are completely determined and match the data distribution, with low-power laptops concentrated in the range of 0~30kW and high-power laptops concentrated in the range of 90~120kW.

[0039] Next, the Gaussian membership function is defined with its center equal to the cluster mean μ and its width equal to the standard deviation σ of the samples within the cluster. The function formula is as follows: Substituting the parameters into the Gaussian membership function in Table 2 above, we obtain... P DR The membership function of each corresponding fuzzy set; The fuzzy set to which each input and output variable in the theoretically optimal energy allocation sequence belongs is determined based on the membership function; continuing with the data in Table 2 as an example, if the power demand in a certain subprocess... P DR =65kW, substituting into the membership functions of each fuzzy set, we can see that: The value is the largest, therefore 65kW is in the M fuzzy set.

[0040] 3. Automatic generation of fuzzy rules based on rule induction Using the fuzzy set of input variables already partitioned in step 2 (e.g., P) DR SL / S / M / H, SOC's L / M / H) and the fuzzy set of output variables (such as P eng Using L / M / H as the standard for defining variable categories, based on the optimal energy allocation database, the total data for each category of the output variable (corresponding to a fuzzy set of the output variable) can be statistically calculated. Then, based on this, the frequency statistics of each category of the input variable (corresponding to the fuzzy set of the input variable) corresponding to the output category are performed.

[0041] The decision tree is rooted at the fuzzy set of the output variables. For example, under each category of fuel cell power, the input variables (such as vehicle power demand) are selected. P DR The fuzzy set of ) is used as the second-level node of the decision tree, and its selection criterion is as follows: P DR The data statistics on the relationship between fuel cell output power and vehicle power demand can be obtained by finding the sample with the highest correlation frequency with the output fuzzy set, or the sample with the smallest variance. Figure 2 The decision tree shown uses P ENG =L is the root node, and we can see the second-level node P. DR The fuzzy set category M has the highest frequency of 846, so it is selected as the target secondary node.

[0042] After selecting the target secondary node, the fuzzy set of the SOC of the power battery is further selected as the third-level node of the decision tree. It can be seen that among the third-level SOC nodes, the fuzzy set category M has the highest frequency of 426, so it is selected as the target third-level node. The mapping relationship between the input fuzzy set combination and the output fuzzy set is extracted from the root node, target secondary node, and target third-level node, and added to the fuzzy rule set, such as... Figure 2 In the process, the extracted fuzzy rule is: if the input variable P DR If the input variable SOC = M, then the output variable P ENG =L.

[0043] The fuzzy rule set refers to data-driven, automatically generated fuzzy rules that can be directly used for energy distribution control in vehicle powertrain systems. An example of the fuzzy rule set is as follows: P DR =SL and SOC=L, then P eng =L; that is, the required power P DR When the battery is in the SL fuzzy set (0-30kW, low power) and the input variable, the power battery SOC, is in the L fuzzy set (0.2-0.4, low charge), the output variable, engine power P, is triggered. eng It is located in the L fuzzy set (0-20kW, low power).

[0044] The most frequent data is chosen as the final rule because it is more likely to capture the rules and relationships within the data, more accurately representing the overall characteristics of the data, thus leading to better optimization results on new data. After each set of rules is learned, all data corresponding to that rule is removed from the sample, and the next rule is extracted and learned from the remaining samples; this process is repeated, extracting trajectories from the optimal dataset until no new rules can be learned.

[0045] One embodiment of this step is as follows: First, extract the fuzzy set labels of the input variables for all samples from the optimal energy allocation database, such as sample P. DR =70kW corresponds to the M fuzzy set, SOC=0.6 corresponds to the M fuzzy set, and the fuzzy set labels for extracting all output variables are shown in Table 3 below: Table 3 For each fuzzy set of the output variable, count the frequency of occurrence of each combination of fuzzy sets of the input variables in all corresponding samples. For example, if 10 samples are traversed, the frequency of occurrence is counted. P eng The number of samples contained in each fuzzy set L / M / H are 3, 4, and 3 respectively. Taking the fuzzy set of the output variable as the core, the frequency of occurrence of the corresponding input variable fuzzy set combinations is counted one by one, as shown in the following example: For Peng 3 samples of =L, statistical analysis P DR The frequency combinations of SOC tags include: Combination 1: P DR The frequency of occurrence of =L and SOC=L is 2; Combination 2: P DR =L and SOC=M occur at a frequency of 1; Combination 3: P DR =S and the frequency of occurrence of SOC=L is 0; for P eng =M's 4 samples, P DR =S and SOC=M occur 1 time; Combination 2: P DR =S and SOC=H occur 1 time; Combination 3: P DR The frequency of occurrence of SOC=M is ​​2; and so on. P eng =H is the frequency of the combination of 3 samples.

[0046] Based on the above statistical results, the generation rules are as follows: P eng When the input combination is equal to L, combination 1 appears most frequently. Therefore, the generation rule is: if the required power... P DR If the power output is SL (0~30kW) and the SOC of the power battery is L (0.2~0.4), then the engine power... P eng For L (0~20kW), the corresponding logic of this rule is: when the power demand is low and the battery is low, the engine does not output power to avoid inefficient operation.

[0047] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0048] This invention also proposes a vehicle power optimization fuzzy control system based on optimal energy management, comprising: Energy Management Optimization Model Construction Module: Used to establish an energy management optimization model that includes power demand, battery status, and energy source constraint parameters based on vehicle operating condition data and vehicle power system model; The optimal energy allocation database generation module is used to divide the energy management process of the vehicle operating cycle into several sub-processes according to the time step based on the objective function minimization criterion of the energy management optimization model and the global optimization method. Following the power demand, battery status and energy source constraint parameters in the energy management optimization model, it enumerates in reverse order to calculate the optimal fuel consumption and electricity consumption schemes for each sub-process and generates the theoretically optimal energy allocation sequence. The fuzzy set and membership function partitioning module is used to perform unsupervised classification of the input and output variables in the theoretically optimal energy allocation sequence using an improved K-means algorithm. This yields fuzzy sets for each input variable and each output variable, including their range, mean, and standard deviation. Table 2 shows the input variable, demand power. P DR The fuzzy set is obtained; a membership function is constructed based on the mean and standard deviation; and the fuzzy set to which each input variable and output variable in the theoretical optimal energy allocation sequence belongs is determined based on the membership function.

[0049] The fuzzy rule generation module is used to count the frequency of the combination of input variable fuzzy sets corresponding to each output variable fuzzy set; taking each fuzzy set classification of the output variable as the root node of the decision tree, selecting two input variable fuzzy sets from the input variables as the second-level and third-level nodes of the decision tree respectively, and filtering out the target second-level node and target third-level node from the second-level and third-level nodes based on the frequency; extracting the mapping relationship between the input variable fuzzy sets and the output variable fuzzy sets from the root node, target second-level node, and target third-level node, traversing each fuzzy set classification of the output variable and the combination of input variables, and finally obtaining the fuzzy rule set composed of the mapping relationship; Control output module: Used to output energy distribution control commands for the vehicle power system according to the fuzzy rule set, and to regulate the output power of the engine / fuel cell and the power battery.

[0050] In one embodiment, the vehicle powertrain model includes an energy source output characteristics and constraint sub-model, a vehicle power demand calculation sub-model, and a transmission system power loss sub-model. The energy source output characteristics and constraint sub-model is used to provide the energy consumption calculation basis and safety constraint boundary for the energy management optimization model. The vehicle power demand calculation sub-model is used to provide the overall power allocation target for the energy management optimization model, and the transmission system power loss sub-model is used to provide power balance constraints for the energy management optimization model.

[0051] In one embodiment, the method for sequentially selecting target second-level nodes and target third-level nodes from second-level nodes and third-level nodes based on the frequency in the fuzzy rule generation module includes: Select an input variable from the input variables, and use the fuzzy set of the input variable as a candidate for the second-level node of the decision tree. Based on the frequency, select the fuzzy set category with the highest combination frequency with the root node from the candidates for the second-level node as the target second-level node. Select one input variable from the remaining input variables, and use the fuzzy set of this input variable as a candidate for a third-level node in the decision tree. Based on the frequency, select the fuzzy set category with the highest combination frequency with the root node from the candidates for third-level nodes as the target third-level node. If there are fuzzy set categories with the same combination frequency, select the fuzzy set category with the smallest sample variance in the corresponding optimal energy allocation database as the target node.

[0052] In one embodiment, the method for traversing each fuzzy set classification of the output variable and the combination of input variables in the fuzzy rule generation module includes: Iterate through each fuzzy set classification of the output variable and the different combinations of all input variables in order; the different combinations of all input variables in order include: all possible permutations of each input variable as a second-level node and other input variables as third-level nodes, ensuring that all combinations of input variables at the second-level and third-level node positions are covered.

[0053] In one embodiment, the method for unsupervised classification of input and output variables in the theoretically optimal energy allocation sequence includes: S1. Randomly select k samples from the dataset of the corresponding variable in the optimal energy allocation database as cluster centers; S2. Calculate the Euclidean distance between each sample in the dataset and all cluster centers, and assign each sample to the cluster containing the nearest cluster center. The range of sample values ​​in each cluster is the initial range of the fuzzy set. S3. Calculate the arithmetic mean of all samples within each cluster to obtain the updated cluster center; S4. If the difference between the updated cluster centers and the previous cluster centers is less than or equal to the set value, then end the current clustering process; otherwise, proceed to step S3. S5. Select the result with the smallest sum of intra-cluster distances from the clustering results as the fuzzy set of each input variable and the fuzzy set of each output variable, including the range of values, mean, and standard deviation.

[0054] In one embodiment, the method for determining the number of clusters k includes: Calculate the sum of Euclidean distances between all samples within a cluster and the cluster center for different k values, plot the curve of the sum of distances as a function of k, and take the k value when the slope of the sum of distances between adjacent k values ​​is less than a set proportion as the number of clusters.

[0055] This invention also provides a fuzzy control method for vehicle power optimization based on optimal energy management, comprising: Based on vehicle operating condition data and vehicle power system model, an energy management optimization model is established that includes power demand, battery status, and energy source constraint parameters. Based on the objective function minimization criterion of the energy management optimization model, the energy management process of the vehicle operating cycle is divided into several sub-processes according to the time step using a global optimization method. Following the power demand, battery status and energy source constraint parameters in the energy management optimization model, the optimal fuel consumption and electricity consumption schemes of each sub-process are calculated in reverse order to generate the theoretically optimal energy allocation sequence. An improved K-means algorithm is used to perform unsupervised classification of the input and output variables in the theoretically optimal energy allocation sequence, resulting in fuzzy sets for each input variable and each output variable, including their value range, mean, and standard deviation. A membership function is constructed based on the mean and standard deviation. The fuzzy set to which each input and output variable in the theoretically optimal energy allocation sequence belongs is determined based on the membership function. The frequency of input variable fuzzy set combinations corresponding to each output variable fuzzy set is counted. Taking each fuzzy set classification of the output variable as the root node of the decision tree, two input variable fuzzy sets are selected from the input variables as the second-level and third-level nodes of the decision tree, respectively. Based on the frequency, the target second-level node and target third-level node are sequentially selected from the second-level and third-level nodes. The mapping relationship between the input variable fuzzy sets and the output variable fuzzy sets is extracted from the root node, the target second-level node, and the target third-level node. Each fuzzy set classification of the output variable and the combination of input variables are traversed to finally obtain the fuzzy rule set composed of the mapping relationship. The energy distribution control command of the vehicle power system is output according to the fuzzy rule set to regulate the output power of the engine / fuel cell and the power battery.

[0056] This invention also provides a computer program product, including a computer program / instructions that, when executed by a processor, implement the various steps of the method described in this invention.

[0057] This invention also provides a non-transitory computer-readable storage medium storing a computer program. The computer program includes program instructions that, when executed by a processor, implement the various steps of the method described in this invention, which will not be elaborated further here.

[0058] The computer-readable storage medium can be the data transmission apparatus or the internal storage unit of a computer device provided in any of the foregoing embodiments, such as the hard disk or memory of the computer device. The computer-readable storage medium can also be the external storage device of the computer device, such as the plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc. equipped on the computer device.

[0059] Furthermore, the computer-readable storage medium may include both internal storage units and external storage devices of the computer device. The computer-readable storage medium is used to store the computer program and other programs and data required by the computer device. The computer-readable storage medium may also be used to temporarily store data that is to be output or has already been output.

[0060] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0061] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0062] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0063] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0064] The contents not described in detail in this specification are existing technologies known to those skilled in the art.

Claims

1. A fuzzy control system for vehicle power optimization based on optimal energy management, characterized in that, include: Energy Management Optimization Model Construction Module: Used to establish an energy management optimization model that includes power demand, battery status, and energy source constraint parameters based on vehicle operating condition data and vehicle power system model; The optimal energy allocation database generation module is used to divide the energy management process of the vehicle operating cycle into several sub-processes according to the time step based on the objective function minimization criterion of the energy management optimization model and the global optimization method. Following the power demand, battery status and energy source constraint parameters in the energy management optimization model, it calculates the optimal fuel consumption and electricity consumption scheme of each sub-process to obtain the theoretical optimal energy allocation sequence. The fuzzy set and membership function partitioning module is used to perform unsupervised classification of the input and output variables in the theoretically optimal energy allocation sequence using an improved K-means algorithm, obtaining fuzzy sets of each input variable and fuzzy sets of each output variable, including their value range, mean, and standard deviation; and constructing membership functions based on the mean and standard deviation. The fuzzy set to which each input and output variable in the theoretically optimal energy allocation sequence belongs is determined based on the membership function. Fuzzy rule generation module: used to count the frequency of combinations of input variable fuzzy sets corresponding to each output variable fuzzy set; Using each fuzzy set classification of the output variable as the root node of the decision tree, two fuzzy sets of input variables are selected from the input variables as the second-level and third-level nodes of the decision tree, respectively. Based on the frequency, the target second-level node and the target third-level node are sequentially selected from the second-level and third-level nodes. The mapping relationship between the fuzzy sets of input variables and the fuzzy sets of output variables is extracted from the root node, the second-level node of the target, and the third-level node of the target. Each fuzzy set classification of the output variable and the combination of input variables are traversed to finally obtain the fuzzy rule set composed of the mapping relationship. Control output module: Used to output energy distribution control commands for the vehicle power system according to the fuzzy rule set, and to regulate the output power of the engine / fuel cell and the power battery.

2. The vehicle power optimization fuzzy control system based on optimal energy management as described in claim 1, characterized in that, The objective function expression of the energy management optimization model includes: In the formula, J represents the total energy consumption cost of the entire operating cycle; T represents the total duration of the operating cycle. P eng ( t () represents the output power of the engine / fuel cell at time t; f ( P eng ( t ))express t Fuel consumption rate of the engine / fuel cell at any given time; P bat ( t () represents the charging / discharging power of the power battery at time t; eP bat ( t )) represents the power battery's energy consumption rate at time t; w 1 and w 2 represents the weighting coefficients used to unify fuel and electricity consumption dimensions.

3. The vehicle power optimization fuzzy control system based on optimal energy management as described in claim 1 or 2, characterized in that, The vehicle powertrain model includes an energy source output characteristics and constraint sub-model, a vehicle power demand calculation sub-model, and a transmission system power loss sub-model; the energy source output characteristics and constraint sub-model is used to provide energy consumption calculation basis and safety constraint boundaries for the energy management optimization model; The vehicle power demand calculation sub-model is used to provide the overall power allocation target for the energy management optimization model, and the power loss sub-model of the transmission system is used to provide power balance constraints for the energy management optimization model.

4. The vehicle power optimization fuzzy control system based on optimal energy management as described in claim 1, characterized in that, In the fuzzy rule generation module, the method for sequentially filtering target second-level nodes and target third-level nodes from second-level nodes and third-level nodes based on the frequency includes: Select an input variable from the input variables, and use the fuzzy set of the input variable as a candidate for the second-level node of the decision tree. Based on the frequency, select the fuzzy set category with the highest combination frequency with the root node from the candidates for the second-level node as the target second-level node. Select one input variable from the remaining input variables, and use the fuzzy set of that input variable as a candidate for the third-level node of the decision tree. Based on the frequency, select the fuzzy set category with the highest combination frequency with the root node from the candidates for the third-level node as the target third-level node.

5. The vehicle power optimization fuzzy control system based on optimal energy management as described in claim 1 or 4, characterized in that, The method for traversing each fuzzy set classification of the output variable and the combination of input variables in the fuzzy rule generation module includes: Iterate through each fuzzy set classification of the output variable and the different combinations of all input variables in order; the different combinations of all input variables in order include: all possible permutations of each input variable as a second-level node and other input variables as third-level nodes, ensuring that all combinations of input variables at the second-level and third-level node positions are covered.

6. The vehicle power optimization fuzzy control system based on optimal energy management as described in claim 1, characterized in that, Unsupervised classification methods for input and output variables in the theoretically optimal energy allocation sequence include: S1. Randomly select k samples from the dataset of the corresponding variable in the optimal energy allocation database as cluster centers; S2. Calculate the Euclidean distance between each sample in the dataset and all cluster centers, and assign each sample to the cluster containing the nearest cluster center. The range of sample values ​​in each cluster is the initial range of the fuzzy set. S3. Calculate the arithmetic mean of all samples within each cluster to obtain the updated cluster center; S4. If the difference between the updated cluster centers and the previous cluster centers is less than or equal to the set value, then end the current clustering process; otherwise, proceed to step S3. S5. Select the result with the smallest sum of intra-cluster distances from the clustering results as the fuzzy set of each input variable and the fuzzy set of each output variable, including the range of values, mean, and standard deviation.

7. The vehicle power optimization fuzzy control system based on optimal energy management as described in claim 6, characterized in that, Methods for determining the number of clusters k include: Calculate the sum of Euclidean distances between all samples within a cluster and the cluster center for different k values, plot the curve of the sum of distances as a function of k, and take the k value when the slope of the sum of distances between adjacent k values ​​is less than a set proportion as the number of clusters.

8. A fuzzy control method for vehicle power optimization based on optimal energy management for the system as described in claim 1, characterized in that, include: Based on vehicle operating condition data and vehicle power system model, an energy management optimization model is established that includes power demand, battery status, and energy source constraint parameters. Based on the objective function minimization criterion of the energy management optimization model, the energy management process of the vehicle operating cycle is divided into several sub-processes according to the time step using a global optimization method. Following the power demand, battery status and energy source constraint parameters in the energy management optimization model, the optimal fuel consumption and electricity consumption schemes for each sub-process are calculated to obtain the theoretically optimal energy allocation sequence. An improved K-means algorithm is used to perform unsupervised classification of the input and output variables in the theoretically optimal energy allocation sequence, resulting in fuzzy sets of each input variable and each output variable, including their value range, mean, and standard deviation. A membership function is then constructed based on the mean and standard deviation. The fuzzy set to which each input and output variable in the theoretically optimal energy allocation sequence belongs is determined based on the membership function. Count the frequency of the combination of input variable fuzzy sets corresponding to each output variable fuzzy set; Using each fuzzy set classification of the output variable as the root node of the decision tree, two fuzzy sets of input variables are selected from the input variables as the second-level and third-level nodes of the decision tree, respectively. Based on the frequency, the target second-level node and the target third-level node are sequentially selected from the second-level and third-level nodes. The mapping relationship between the fuzzy sets of input variables and the fuzzy sets of output variables is extracted from the root node, the second-level node of the target, and the third-level node of the target. Each fuzzy set classification of the output variable and the combination of input variables are traversed to finally obtain the fuzzy rule set composed of the mapping relationship. The energy distribution control command of the vehicle power system is output according to the fuzzy rule set to regulate the output power of the engine / fuel cell and the power battery.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the vehicle power optimization fuzzy control method based on optimal energy management as described in claim 8.

10. A computer program product comprising a computer program / instructions, characterized in that, When the computer program / instruction is executed by the processor, it implements the steps of the vehicle power optimization fuzzy control method based on optimal energy management as described in claim 8.

Images