Pantograph type heavy truck battery and power grid power cooperative control method and system

Abstract

The invention provides a pantograph type heavy truck battery and power grid power cooperative control method and system, and relates to the field of power supply, and the method comprises the steps: constructing a heterogeneous state feature vector based on the state data and demand power of a vehicle; generating a battery output power reference instruction at the current moment according to the heterogeneous state feature vector through a strategy network based on a dual-channel feature decoupling mechanism; engineering constraint verification and cutting are carried out on the battery output power reference instruction, and the battery output power at the current moment is generated; and calculating the power grid side output power based on the battery-power grid power coupling relationship, the battery output power and the demand power at the current moment, and the method has the advantages that the traction power response performance is ensured, the sudden change of the power grid side output power is inhibited, and the cooperative control effect of long-life operation of the battery and smooth output of the power grid power is realized.

Description

Technical Field

[0001] This invention relates to the field of power supply, and in particular to a method and system for coordinated control of pantograph-type heavy truck batteries and grid power. Background Technology

[0002] Electrified highway systems are an important way for heavy-duty vehicles to reduce fuel consumption and carbon emissions. Pantograph-equipped heavy-duty trucks draw power from both the onboard pantograph and the roadside overhead contact line, forming a dual-power source system with the power battery. During operation, the power output from the battery and the grid needs to be dynamically allocated in real time. However, pantograph-equipped heavy-duty trucks face the dual challenges of stringent constraints from transient power surges on the grid side and ensuring the lifespan of the power battery.

[0003] Current research on power allocation between batteries and the power grid primarily employs rule-based control, model predictive control, or static optimization methods, but these approaches have significant shortcomings. Firstly, existing methods treat battery state (e.g., SOC) and transient driving demands (e.g., traction power) as a single, flattened input, failing to consider their heterogeneous coupling across time scales and physical dimensions. Vehicle external traction demands are influenced by various factors such as road gradient and vehicle speed, exhibiting high-frequency, highly stochastic dynamic changes; while battery state has cumulative and slowly varying characteristics. These two differ significantly in time scale and physical properties, making this approach inaccurate in reflecting the actual situation and affecting the accuracy of power allocation.

[0004] On the other hand, regarding grid stability assurance, existing power allocation methods primarily aim to meet the instantaneous traction needs of vehicles, lacking an endogenous constraint mechanism for the dynamic response characteristics of the overhead contact line power supply system. In actual operation, when operating conditions change abruptly, the output power on the grid side will fluctuate significantly. This fluctuation can easily cause current surges in roadside converters and the overhead contact line, thereby affecting the safety and reliability of the power supply system. For example, when vehicles suddenly accelerate or decelerate, existing methods cannot adjust power allocation in a timely and effective manner, resulting in an unstable transition of output power on the grid side and increasing the risk of power supply system failure.

[0005] Therefore, there is a need to provide a method and system for coordinated control of pantograph-type heavy truck batteries and grid power, in order to suppress sudden changes in grid output power while ensuring traction power response performance, and achieve a coordinated control effect of long battery life operation and smooth grid power output. Summary of the Invention

[0006] This invention provides a method and system for coordinated control of battery and grid power in pantograph-type heavy-duty trucks, comprising: constructing a heterogeneous state feature vector based on vehicle state data and power demand; generating a battery output power reference command at the current moment based on the heterogeneous state feature vector through a strategy network based on a dual-channel feature decoupling mechanism; performing engineering constraint verification and trimming on the battery output power reference command to generate the battery output power at the current moment; and calculating the grid-side output power based on the battery-grid power coupling relationship, the battery output power, and the power demand at the current moment.

[0007] Furthermore, based on the vehicle's state data and power demand, a heterogeneous state feature vector is constructed, including: constructing internal battery energy features corresponding to multiple moments based on the battery SOC value, the battery output power, battery voltage, and battery current corresponding to multiple moments in the vehicle's state data; and constructing external traction features corresponding to multiple moments based on the speed and power demand corresponding to multiple moments in the vehicle's state data, wherein the heterogeneous state feature vector includes internal battery energy features and external traction features corresponding to multiple moments.

[0008] Furthermore, the policy network based on the dual-channel feature decoupling mechanism includes an external load channel, an internal energy channel, a decoupling constraint module, a feature fusion module, and a policy generation module. The external load channel is used to construct an external load input sequence based on external traction features corresponding to multiple time points, and to extract external load features from the external load input sequence. The internal energy channel is used to construct an internal energy input sequence based on internal battery energy features corresponding to multiple time points, and to extract internal energy features from the internal energy input sequence. The decoupling constraint module is used to decouple the external load features and internal energy features, generating decoupled external load features and internal energy features. The feature fusion module is used to fuse the decoupled external load features and internal energy features to generate fused features. The policy generation module is used to generate a battery output power reference command based on the fused features.

[0009] Furthermore, the external load channel includes an input sequence construction module, a feature extraction module, a feature compression and mapping module, and a stability enhancement module. The input sequence construction module is used to construct an external load input sequence based on external traction features corresponding to multiple time points. The feature extraction module is used to extract external load dynamic characteristics from the external load input sequence through a temporal convolutional network or a gated recurrent unit. The feature compression and mapping module is used to generate initial external load features based on the external load dynamic characteristics. The stability enhancement module is used to generate external load features based on the initial external load features, wherein the stability enhancement module includes at least a normalization layer and a Dropout layer.

[0010] Furthermore, the decoupling constraint module decouples the external load characteristics and internal energy characteristics to generate decoupled external load characteristics and internal energy characteristics, including: decoupling the external load characteristics and internal energy characteristics by reducing the correlation and / or orthogonalizing the external load characteristics and internal energy characteristics to generate decoupled external load characteristics and internal energy characteristics.

[0011] Furthermore, the feature fusion module performs feature fusion on the decoupled external load features and internal energy features to generate fused features, including: splicing the decoupled external load features and internal energy features and generating fusion weights; and performing feature fusion on the decoupled external load features and internal energy features through the fusion weights to generate fused features.

[0012] Furthermore, engineering constraint verification and trimming are performed on the battery output power reference command to generate the battery output power, including: performing engineering constraint verification and trimming on the battery output power reference command according to the constraint set to generate the battery output power, wherein the constraint set includes battery charging and discharging power boundary constraints, grid power supply capacity constraints, battery-grid power change constraints, and SOC safety constraints.

[0013] Furthermore, based on the battery-grid power coupling relationship, battery output power, and current demand power, the grid-side output power is calculated, including: calculating the initial grid-side output power based on the battery-grid power coupling relationship, battery output power, and current demand power; determining whether the initial grid-side output power meets the grid power constraint conditions; if so, using the initial grid-side output power as the grid-side output power; if not, performing secondary compensation on the battery output power based on the grid power constraint conditions and the initial grid-side output power; and determining whether to adjust the current demand power based on the secondary compensated battery output power and the battery output power constraint conditions; if not, generating the grid-side output power based on the battery-grid power coupling relationship, the secondary compensated battery output power, and current demand power; if so, correcting the current demand power based on the grid power constraint conditions and the battery output power constraint conditions.

[0014] Furthermore, based on the grid power constraints and the initial grid-side output power, the battery output power is compensated a second time, including: determining the power compensation value based on the initial grid-side output power and the maximum grid-side output power; and generating the battery output power after secondary compensation based on the battery output power and the power compensation value.

[0015] This invention provides a pantograph-type heavy-duty truck battery and grid power coordinated control system, comprising: a data processing module for constructing a heterogeneous state feature vector based on vehicle state data and power demand; an instruction generation module for generating a battery output power reference instruction at the current moment based on the heterogeneous state feature vector through a strategy network based on a dual-channel feature decoupling mechanism; a power generation module for performing engineering constraint verification and trimming on the battery output power reference instruction to generate the battery output power at the current moment; and a power calculation module for calculating the grid-side output power based on the battery-grid power coupling relationship, the battery output power, and the power demand at the current moment.

[0016] Compared with existing technologies, the pantograph-type heavy truck battery and grid power coordinated control method and system provided by the present invention have at least the following beneficial effects: 1. A strategy network is constructed based on a dual-channel feature decoupling mechanism to process internal battery energy features and external traction features separately. The decoupling constraint module reduces the correlation between the two types of features, eliminates mutual interference, and makes feature extraction more accurate. The feature fusion module dynamically generates fusion weights according to different operating conditions, reasonably adjusting the relative importance of the two types of features in the decision-making process, enabling the strategy generation module to generate battery output power reference commands based on comprehensive and accurate information. 2. The reference command for battery output power undergoes engineering constraint verification and trimming. This involves comprehensively considering constraints such as battery charging / discharging power boundaries, grid power supply capacity, power variations, and SOC safety to ensure the battery output power remains within a safe and reasonable range, preventing damage to the battery and grid due to abnormal power. When calculating the grid-side output power, secondary compensation and demand power correction are performed by determining whether the initial value meets grid power constraints. This ensures the grid-side output power meets requirements, maintains the stability of battery and grid power coordination control, and guarantees the safe and reliable operation of the entire pantograph-type heavy-duty truck system. 3. Based on the vehicle's real-time status and demands, rationally allocate the output power of the battery and the power grid. While meeting the vehicle's power requirements, fully consider the actual conditions of the battery and the power grid, and optimize the power allocation strategy through mechanisms such as secondary compensation and demand power correction to improve energy utilization efficiency. This avoids overcharging and discharging of the battery, extending its lifespan, while fully utilizing the power grid's supply capacity, reducing energy waste, and achieving efficient coordinated control of the pantograph-type heavy-duty truck battery and grid power, thereby improving the overall energy utilization level. Attached Figure Description

[0017] This specification will be further described by way of exemplary embodiments, which will be described in detail with reference to the accompanying drawings. These embodiments are not limiting; in these embodiments, the same reference numerals denote the same structures, wherein: Figure 1This is a schematic flowchart illustrating a pantograph-type heavy truck battery and grid power coordinated control method according to some embodiments of this specification; Figure 2 This is a schematic diagram of a strategy network based on a dual-channel feature decoupling mechanism, as shown in some embodiments of this specification. Figure 3(a) is a schematic diagram of the speed of driving cycle 1 according to some embodiments of this specification; Figure 3(b) is a schematic diagram of the power demand for driving cycle 1 according to some embodiments of this specification; Figure 4(a) is a schematic diagram of the speed of driving cycle 2 according to some embodiments of this specification; Figure 4(b) is a schematic diagram of the power demand for driving cycle 2 according to some embodiments of this specification; Figure 5(a) is a schematic diagram of the speed of driving cycle 3 according to some embodiments of this specification; Figure 5(b) is a schematic diagram of the power demand for driving cycle 3 according to some embodiments of this specification; Figure 6 This is a schematic diagram of the battery-grid power coordinated control results corresponding to driving cycle 1 as shown in some embodiments of this specification; Figure 7 This is a schematic diagram of the battery-grid power coordinated control results corresponding to driving cycle 2 as shown in some embodiments of this specification; Figure 8 This is a schematic diagram of the battery-grid power coordinated control results corresponding to driving cycle 3 as shown in some embodiments of this specification; Figure 9 This is a schematic diagram illustrating the decision-making time according to some embodiments of this specification; Figure 10 This is a schematic diagram of a pantograph-type heavy truck battery and grid power coordinated control system according to some embodiments of this specification. Detailed Implementation

[0018] To more clearly illustrate the technical solutions of the embodiments in this specification, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are merely some examples or embodiments of this specification. For those skilled in the art, these drawings can be applied to other similar scenarios without creative effort. Unless obvious from the context or otherwise specified, the same reference numerals in the drawings represent the same structures or operations.

[0019] Figure 1 This is a flowchart illustrating a pantograph-type heavy-duty truck battery and grid power coordinated control method according to some embodiments of this specification, such as... Figure 1 As shown, the method for coordinated control of pantograph-type heavy truck battery and grid power may include the following steps.

[0020] Step 110: Construct heterogeneous state feature vectors based on vehicle state data and power demand.

[0021] Specifically, for each moment, the collection of vehicle status data and required power may include: the controller of the pantograph-type heavy truck collects the accelerator pedal opening, brake pedal opening, vehicle speed, and gear signal in real time; the torque management module calculates the appropriate drive or braking torque; and after low-pass filtering, the required power is finally obtained. Meanwhile, the vehicle's status data also includes: battery state of charge. Battery voltage and current, etc.

[0022] In some embodiments, step 110 specifically includes: Based on the battery SOC value at multiple times in the vehicle's state data, the battery output power at the previous time, the battery voltage and the battery current, the internal battery energy characteristics at multiple times are constructed. Based on the speed and power demand at multiple moments in the vehicle's state data, external traction features at multiple moments are constructed. The heterogeneous state feature vector includes internal battery energy features and external traction features at multiple moments.

[0023] Specifically, the internal battery energy characteristics and external traction characteristics at a certain moment can be expressed as follows: in, The internal battery energy characteristics at time k are shown below. The external traction feature corresponding to the k-th time point. Let SOC be the battery value at time k. It represents the action at time k-1, indicating the battery's output power at time k-1. These are the battery voltage and current at time k. These are the power and speed required for the current vehicle operation.

[0024] Step 120: Using a strategy network based on a dual-channel feature decoupling mechanism, the battery output power reference command for the current moment is generated according to the heterogeneous state feature vector.

[0025] Figure 2 This is a schematic diagram of a policy network based on a dual-channel feature decoupling mechanism, as shown in some embodiments of this specification. Figure 2As shown, in some embodiments, the policy network based on the dual-channel feature decoupling mechanism includes an external load channel, an internal energy channel, a decoupling constraint module, and a feature fusion module. The external load channel is used to construct an external load input sequence based on the external traction characteristics corresponding to multiple time points, and to extract external load characteristics from the external load input sequence; The internal energy channel is used to construct an internal energy input sequence based on the internal battery energy characteristics corresponding to multiple time points, and to extract internal energy characteristics from the internal energy input sequence. The decoupling constraint module is used to decouple external load characteristics and internal energy characteristics, generating decoupled external load characteristics and internal energy characteristics; The feature fusion module is used to fuse the decoupled external load features and internal energy features to generate fused features; The policy network based on the dual-channel feature decoupling mechanism also includes a policy generation module, which is used to generate battery output power reference commands based on fused features.

[0026] In some embodiments, the external load channel includes an input sequence construction module, a feature extraction module, a feature compression and mapping module, and a stability enhancement module; The input sequence construction module is used to construct the external load input sequence based on the external traction characteristics corresponding to multiple time points; The feature extraction module is used to extract the dynamic characteristics of the external load from the external load input sequence through a temporal convolutional network or a gated recurrent unit. The feature compression and mapping module is used to generate initial external load features based on the dynamic characteristics of the external load; The stability enhancement module is used to generate external load characteristics based on the initial external load characteristics. The stability enhancement module includes at least a normalization layer and a Dropout layer.

[0027] Specifically, the input sequence construction module is used to orderly combine the external traction features corresponding to multiple moments in chronological order over a period of time, thereby constructing an input sequence that can comprehensively reflect the changes in external load over a period of time, providing rich and accurate basic data for subsequent feature extraction.

[0028] The feature extraction module extracts valuable dynamic characteristics of the external load from the constructed external load input sequence. This module employs a temporal convolutional network (TCNN) to achieve this function. The TCNN captures and analyzes local information in the input sequence through convolution operations. When processing the external load input sequence, the TCNN can identify the changing patterns of traction power demand and vehicle speed at different times, extracting short-term dynamic characteristics of load demand, such as large fluctuations and periodic changes in load over a short period, laying the foundation for generating accurate external load features. In some other implementations, a gated recurrent unit (GRU) can also be used to perform feature extraction, effectively processing the external load input sequence and capturing long-term dependencies within it. The TCNN includes at least two one-dimensional convolutional layers, each with a kernel size of 3, 32 convolutional channels, and dilation coefficients set to 1 and 2 respectively; the GRU adopts a single-layer structure with a hidden state dimension of 32.

[0029] The feature compression and mapping module generates initial external load features based on the external load dynamic characteristics obtained by the feature extraction module. Since the extracted dynamic characteristics may contain a large amount of redundant information and have high dimensionality, which is not conducive to subsequent processing and analysis, this module compresses these dynamic characteristics to remove unnecessary redundant information. Simultaneously, through a specific mapping relationship, it transforms the processed data into a suitable feature space, generating initial external load features with a more reasonable dimensionality, enabling a more concise and effective representation of relevant external load information.

[0030] The stability enhancement module improves the numerical stability and robustness of the feature extraction process, generating reliable final external load features based on the initial external load features. This module includes at least a normalization layer and a Dropout layer. The normalization layer standardizes the values ​​of the initial external load features, adjusting feature data of different scales to a uniform range, avoiding numerical instability caused by excessive differences in data scale, and making subsequent calculations more accurate and stable. The Dropout layer randomly discards a portion of neurons during training to prevent overfitting, enhance the model's adaptability to different data, improve the robustness of feature extraction, and thus generate high-quality, stable external load features. .

[0031] In some embodiments, the internal energy channel may include an input sequence construction module, a smoothing preprocessing module, a gated loop unit, a feature compression and mapping module, and a stability enhancement module. The input sequence construction module is used to construct an internal energy input sequence by sequentially combining internal battery energy features corresponding to multiple time points over a time span. The smoothing preprocessing module primarily addresses transient fluctuations in the internal energy input sequence. In actual operation, due to various interference factors, the data in the internal energy input sequence may experience brief and drastic changes, which can affect the accuracy of subsequent feature extraction. The smoothing preprocessing module uses appropriate filtering algorithms, such as moving average filtering and exponential smoothing filtering, to smooth the input sequence, suppressing these transient fluctuations and making the data more stable.

[0032] The gated recurrent unit (GRU) is used to extract the long-term evolution characteristics of the energy state, capturing the long-term dependencies between data at different times in the input sequence. When processing the input sequence of the internal energy channel, the GRU can analyze the changing trends of variables such as battery SOC, voltage, and current over time, uncovering the long-term evolution patterns of the energy state.

[0033] The feature compression and mapping module further generates internal energy features based on the long-term evolution features extracted from the gated recurrent unit. This module compresses the extracted features, removes redundant information, and transforms the features into a suitable feature space through a specific mapping relationship, generating internal energy features with reasonable dimensions and greater representativeness.

[0034] The stability enhancement module is used to improve the temporal continuity and numerical stability of internal energy characteristics. It ensures good continuity of internal energy characteristics across different time points by employing normalization and adding noise, avoiding discontinuities caused by data abrupt changes. Simultaneously, it improves the numerical stability of the characteristics, enabling them to more accurately reflect the battery's internal energy state.

[0035] In some embodiments, the decoupling constraint module decouples external load characteristics and internal energy characteristics to generate decoupled external load characteristics and internal energy characteristics, including: By reducing the correlation between external load characteristics and internal energy characteristics and / or performing orthogonalization operations, the external load characteristics and internal energy characteristics are decoupled, generating decoupled external load characteristics and internal energy characteristics.

[0036] Specifically, during vehicle operation, rapid fluctuations in external loads, such as sudden changes in traction due to abrupt road conditions, can be transmitted to internal energy characteristics through coupling mechanisms, affecting the stable operation of internal systems such as battery energy management. By calculating correlation coefficients and mutual information between the two types of characteristics, the degree of their correlation is quantified. Then, dimensionality reduction methods such as principal component analysis are used to recombine and transform the characteristics while retaining the main feature information. This significantly reduces the correlation between the transformed external load characteristics and internal energy characteristics. In this way, changes in external loads will not excessively affect internal energy characteristics, thus achieving preliminary decoupling.

[0037] Orthogonalization, from the perspective of vector space, treats external load features and internal energy features as vectors in a vector space. Through mathematical methods such as Gram-Schmidt orthogonalization and singular value decomposition-based orthogonalization, these two types of feature vectors are processed to make them pairwise orthogonal in the vector space. Orthogonality means that the angle between two vectors is 90 degrees, they are completely independent in direction, and there is no linear dependence. After orthogonalization, the external load features and internal energy features are perpendicular to each other in the vector space, with no overlapping information components, thus fundamentally achieving deep decoupling.

[0038] For example, the projection components of the external load feature vector onto the internal energy feature direction are first eliminated to obtain the orthogonalized decoupling features: in, The external load characteristics after decoupling. It is the identity matrix. This is the regularization parameter.

[0039] In some embodiments, the feature fusion module performs feature fusion on the decoupled external load features and internal energy features to generate fused features, including: The decoupled external load characteristics and internal energy characteristics are spliced ​​together to generate fusion weights; By using fusion weights, the decoupled external load characteristics and internal energy characteristics are fused to generate fused features.

[0040] Specifically, the feature fusion module first uses a concatenation operation to sequentially connect the decoupled external load features and internal energy features. This operation is simple and direct, integrating the originally independent external load and internal energy feature information into a new feature vector. The concatenated feature vector is then input into a fully connected layer. The fully connected layer has powerful feature transformation capabilities; it performs a linear transformation on the input feature vector and automatically adjusts the weight relationships between features by learning from a large number of data samples, uncovering deeper feature combination patterns. After processing by the fully connected layer, the feature vector is activated using the sigmoid function. The sigmoid function strictly limits the output value to the interval (0,1), thereby generating fusion weights. These fusion weights play a crucial role in dynamic adjustment, flexibly changing the relative importance of external load features and internal energy features in the decision-making process according to different operating conditions and requirements. For example, when a vehicle is traveling at high speed and under heavy load, external load characteristics may be more critical, and the fusion weights will assign greater weight to external load characteristics. Conversely, when the battery charge is low and energy needs to be allocated appropriately, the weight of internal energy characteristics will increase accordingly. Finally, through a weighted approach, the decoupled external load characteristics and internal energy characteristics are fused based on the fusion weights to generate fused features. in, As a feature of fusion, To integrate weights, This represents the internal energy characteristics after decoupling.

[0041] The policy generation module uses a multilayer perceptron to map the fused features into control actions at the current moment. This enables an adaptive balance between traction demand response performance and battery energy safety constraints under different operating conditions. The value range is [-1, 1], and the corresponding battery output power reference command can be obtained through linear mapping. .

[0042] Step 130: Perform engineering constraint verification and trimming on the battery output power reference command to generate the battery output power at the current moment.

[0043] Specifically, it includes: Based on the constraint set, the engineering constraint verification and trimming of the battery output power reference command are performed to generate the battery output power. The constraint set includes battery charging and discharging power boundary constraints, grid power supply capacity constraints, battery-grid power change constraints, and SOC safety constraints.

[0044] Specifically, battery charge and discharge power boundary constraints include determining the upper and lower limits of allowable charge and discharge power based on the battery's operating state. The following formula can be used to perform engineering constraint verification and trimming on the battery output power reference command based on the battery charge and discharge power boundary constraints: Among them, the maximum charging and discharging power and minimum charge / discharge power Based on the open-circuit voltage model and combined with the battery SOC value ( ), battery voltage Optional parameters such as temperature can be dynamically determined; alternatively, they can be obtained using a lookup table method or a linear interpolation method.

[0045] The following formula can be used to perform engineering constraint verification and trimming on the battery output power reference command based on grid power supply capacity constraints: in, This is the maximum allowable power supply for the electrified highway power supply system under current operating conditions, in order to avoid transient impacts on the overhead contact line and roadside converters. The power demand at the current moment, This refers to the output power on the grid side.

[0046] Battery-grid power variation constraints; to suppress sudden power changes between adjacent control cycles and limit the rate of power change between the battery side and the grid side, engineering constraint verification and trimming can be performed on the battery output power reference command according to the following formula: in, These represent the upper limits of battery power variation and grid power variation, respectively, thereby improving the smoothness of power allocation commands in the time domain. It represents the actual output power of the battery and the power grid at time k-1. T It is a single control cycle, typically less than 100ms.

[0047] SOC safety constraints; to ensure the battery operates within a safe range, the following conditions must be met: in, These represent the upper and lower limits of the battery's State of Charge (SOC), respectively. Through the above engineering constraint verification and trimming processes, the battery output power that meets the requirements of battery safety, grid stability, and real-time control is obtained. And use it as the execution control quantity of the vehicle's power system.

[0048] Step 140: Calculate the grid-side output power based on the battery-grid power coupling relationship, battery output power, and current power demand.

[0049] Specifically, it includes: Based on the battery-grid power coupling relationship, battery output power and current power demand, calculate the initial grid-side output power; Determine whether the initial grid-side output power meets the grid power constraint condition. If so, use the initial grid-side output power as the grid-side output power. Specifically, determine whether the initial grid-side output power is greater than the maximum grid-side output power. If not, use the initial grid-side output power as the grid-side output power. If not, then the battery output power is compensated a second time based on the grid power constraint and the initial grid-side output power; Based on the battery output power after secondary compensation and the battery output power constraints, it is determined whether to adjust the current demand power. Specifically, it is determined whether the battery output power after secondary compensation is greater than the maximum battery output power. If not, the grid-side output power is generated based on the battery-grid power coupling relationship, the battery output power after secondary compensation, and the current demand power. If so, the current demand power is corrected based on the grid power constraints and battery output power constraints. ,in, The corrected power demand at the current moment. This represents the maximum output power of the battery. This represents the maximum output power on the grid side. At this point, the output power on the grid side is the maximum output power on the grid side, and the output power on the battery is the maximum output power on the battery side.

[0050] Specifically, the initial grid-side output power can be calculated based on the following formula: in, This represents the initial grid-side output power. This represents the battery output power at the current moment.

[0051] In some embodiments, secondary compensation is performed on the battery output power based on grid power constraints and the initial grid-side output power, including: The power compensation value is determined based on the initial grid-side output power and the maximum grid-side output power. Based on the battery output power and power compensation value, the battery output power after secondary compensation is generated.

[0052] Specifically, the battery output power after secondary compensation can be calculated using the following formula: in, This is the power compensation value. For the maximum grid-side output power, This represents the battery output power after secondary compensation.

[0053] The following section uses specific experimental data to illustrate the beneficial effects of the pantograph-type heavy truck battery and grid power coordinated control method.

[0054] Taking a certain model of pantograph-type heavy truck as an example, simulation verification was conducted. The dual-delay deep deterministic policy gradient (TD3) algorithm was used to train the policy network based on a dual-channel feature decoupling mechanism. A value evaluation mechanism was introduced during the training phase to evaluate the policy output results, which can effectively reduce the impact of estimation bias on the policy convergence stability during training. The value evaluation mechanism can be composed of multiple structurally consistent and parameter-independent evaluation units. Its specific implementation is not limited to a specific algorithm form, and those skilled in the art can configure it according to computing power conditions and application requirements. To ensure the comprehensiveness of the experiment and the reliability of the results, three different driving cycles were used as verification data, as shown in Figures 3-5.

[0055] The vehicle is equipped with a lithium iron phosphate battery with a nominal capacity of 170Ah and a rated voltage of 350V. The battery charging and discharging power is limited to [specific parameters]. The SOC operating range is [0.2, 0.8], with an initial SOC value of 0.65. Control cycle. Maximum input power of the power grid .

[0056] The policy network based on the dual-channel feature decoupling mechanism was trained offline, and its hyperparameters are shown in Table 1.

[0057] Table 1 Figure 6 This is a schematic diagram of the battery-grid power coordinated control results corresponding to driving cycle 1, as shown in some embodiments of this specification. Figure 7 This is a schematic diagram of the battery-grid power coordinated control results corresponding to driving cycle 2, as shown in some embodiments of this specification. Figure 8 This is a schematic diagram of the battery-grid power coordinated control results corresponding to driving cycle 3, as shown in some embodiments of this specification, such as... Figures 6-8 As shown, the left side is a graph showing the changes in demand power (black line), battery output power (blue line), and grid-side output power (red line) for driving cycle 1, driving cycle 2, and driving cycle 3, respectively. The right side is a schematic diagram showing the changes in battery SOC for driving cycle 1, driving cycle 2, and driving cycle 3, respectively.

[0058] The experimental results of different algorithms are shown in Table 2.

[0059] Table 2 based on Figures 6-8 As shown in Table 2, this method exhibits significant advantages across all three driving cycles: In terms of energy consumption, this invention reduces energy consumption by 5%-15% compared to the rule-based algorithm, reaching a minimum of 0.72 kWh in driving cycle 3, demonstrating its high-efficiency energy consumption optimization advantage; in terms of grid-side power, this invention controls its variance within the range of 67.7-99.4, significantly better than the other three methods, effectively suppressing transient impacts and improving grid stability; in terms of battery-side power, this invention further optimizes the variance to 27.0-66.3, which is up to 16.9% lower than single-channel TD3, reducing the burden of high-rate re-discharge of the battery; simultaneously, it minimizes SOC fluctuations, ensuring that the battery operates within a safe range and extending its service life.

[0060] Figure 9 This is a schematic diagram illustrating the decision-making time according to some embodiments of this specification, such as... Figure 9 As shown, under different operating conditions, this method can complete the decision output and engineering constraint verification within a single control cycle, meeting the millisecond-level control requirements of heavy trucks on electrified highways, and verifying the feasibility and stability of this method in actual engineering application scenarios.

[0061] Figure 10 This is a schematic diagram of a pantograph-type heavy-duty truck battery and grid power coordinated control system according to some embodiments of this specification, such as... Figure 10 As shown, the pantograph-type heavy truck battery and grid power coordinated control system may include a data processing module, an instruction generation module, a power generation module, and a power calculation module.

[0062] The data processing module is used to construct heterogeneous state feature vectors based on vehicle state data and power demand. The instruction generation module is used to generate the battery output power reference instruction at the current moment based on the heterogeneous state feature vector through a policy network based on a dual-channel feature decoupling mechanism. The power generation module is used to perform engineering constraint verification and trimming on the battery output power reference command to generate the battery output power at the current moment. The power calculation module is used to calculate the grid-side output power based on the battery-grid power coupling relationship, battery output power, and current power demand.

[0063] Finally, it should be understood that the embodiments described in this specification are merely illustrative of the principles of the embodiments described herein. Other variations may also fall within the scope of this specification. Therefore, alternative configurations of the embodiments described herein are intended to be illustrative rather than limiting, and should be considered consistent with the teachings of this specification. Accordingly, the embodiments described herein are not limited to those explicitly introduced and described herein.

Claims

1. A method for coordinated control of battery and grid power in pantograph-type heavy-duty trucks, characterized in that, include: Based on vehicle state data and power demand, a heterogeneous state feature vector is constructed. The battery output power reference command at the current moment is generated by a policy network based on a dual-channel feature decoupling mechanism and heterogeneous state feature vectors. Perform engineering constraint verification and trimming on the battery output power reference command to generate the battery output power at the current moment; The grid-side output power is calculated based on the battery-grid power coupling relationship, battery output power, and current power demand.

2. The pantograph-type heavy truck battery and grid power coordinated control method according to claim 1, characterized in that, Based on vehicle state data and power demand, a heterogeneous state feature vector is constructed, including: Based on the battery SOC value at multiple times in the vehicle's state data, the battery output power at the previous time, the battery voltage and the battery current, the internal battery energy characteristics at multiple times are constructed. Based on the speed and power demand at multiple moments in the vehicle's state data, external traction features corresponding to multiple moments are constructed. The heterogeneous state feature vector includes internal battery energy features and external traction features corresponding to multiple moments.

3. The pantograph-type heavy truck battery and grid power coordinated control method according to claim 2, characterized in that, The policy network based on the dual-channel feature decoupling mechanism includes an external load channel, an internal energy channel, a decoupling constraint module, a feature fusion module, and a policy generation module. The external load channel is used to construct an external load input sequence based on the external traction characteristics corresponding to multiple times, and to extract external load characteristics from the external load input sequence; The internal energy channel is used to construct an internal energy input sequence based on the internal battery energy characteristics corresponding to multiple times, and to extract internal energy characteristics from the internal energy input sequence. The decoupling constraint module is used to decouple the external load characteristics and internal energy characteristics, and generate decoupled external load characteristics and internal energy characteristics. The feature fusion module is used to fuse the decoupled external load features and internal energy features to generate fused features. The strategy generation module is used to generate a battery output power reference command based on the fusion features.

4. The pantograph-type heavy truck battery and grid power coordinated control method according to claim 3, characterized in that, The external load channel includes an input sequence construction module, a feature extraction module, a feature compression and mapping module, and a stability enhancement module; The input sequence construction module is used to construct an external load input sequence based on the external traction characteristics corresponding to multiple time points; The feature extraction module is used to extract the dynamic characteristics of the external load from the external load input sequence through a temporal convolutional network or a gated recurrent unit. The feature compression and mapping module is used to generate initial external load features based on the dynamic characteristics of the external load. The stability enhancement module is used to generate external load characteristics based on the initial external load characteristics, wherein the stability enhancement module includes at least a normalization layer and a Dropout layer.

5. The pantograph-type heavy truck battery and grid power coordinated control method according to claim 4, characterized in that, The decoupling constraint module decouples external load characteristics and internal energy characteristics, generating decoupled external load characteristics and internal energy characteristics, including: By reducing the correlation between external load characteristics and internal energy characteristics and / or performing orthogonalization operations, the external load characteristics and internal energy characteristics are decoupled, generating decoupled external load characteristics and internal energy characteristics.

6. The pantograph-type heavy truck battery and grid power coordinated control method according to claim 4, characterized in that, The feature fusion module performs feature fusion on the decoupled external load features and internal energy features to generate fused features, including: The decoupled external load characteristics and internal energy characteristics are spliced ​​together to generate fusion weights; By using fusion weights, the decoupled external load characteristics and internal energy characteristics are fused to generate fused features.

7. The pantograph-type heavy truck battery and grid power coordinated control method according to any one of claims 1-6, characterized in that, Perform engineering constraint verification and trimming on the battery output power reference command to generate the battery output power, including: Based on the constraint set, the engineering constraint verification and trimming of the battery output power reference command are performed to generate the battery output power. The constraint set includes battery charging and discharging power boundary constraints, grid power supply capacity constraints, battery-grid power change constraints, and SOC safety constraints.

8. The pantograph-type heavy truck battery and grid power coordinated control method according to any one of claims 1-6, characterized in that, Based on the battery-grid power coupling relationship, battery output power, and current power demand, the grid-side output power is calculated, including: Based on the battery-grid power coupling relationship, battery output power and current power demand, calculate the initial grid-side output power; Determine whether the initial grid-side output power meets the grid power constraint conditions. If so, use the initial grid-side output power as the grid-side output power. If not, then the battery output power is compensated twice based on the grid power constraints and the initial grid-side output power; Based on the battery output power after secondary compensation and the battery output power constraint, it is determined whether to adjust the current demand power. If not, the grid-side output power is generated based on the battery-grid power coupling relationship, the battery output power after secondary compensation, and the current demand power. If yes, the current demand power is corrected based on the grid power constraint and the battery output power constraint.

9. The pantograph-type heavy truck battery and grid power coordinated control method according to claim 8, characterized in that, Based on the grid power constraints and the initial grid-side output power, secondary compensation is performed on the battery output power, including: The power compensation value is determined based on the initial grid-side output power and the maximum grid-side output power. Based on the battery output power and power compensation value, the battery output power after secondary compensation is generated.

10. A pantograph-type heavy-duty truck battery and grid power coordinated control system, characterized in that, The method for coordinating the control of battery and grid power in a pantograph-type heavy-duty truck according to any one of claims 1-9 includes: The data processing module is used to construct heterogeneous state feature vectors based on vehicle state data and power demand. The instruction generation module is used to generate the battery output power reference instruction at the current moment based on the heterogeneous state feature vector through a policy network based on a dual-channel feature decoupling mechanism. The power generation module is used to perform engineering constraint verification and trimming on the battery output power reference command to generate the battery output power at the current moment. The power calculation module is used to calculate the grid-side output power based on the battery-grid power coupling relationship, battery output power, and current power demand.

Images