A control method and system for a vehicle brake fluid filling process
By generating multi-dimensional protocol feature fingerprints and feature association network deconstruction, the brake fluid filling equipment has achieved autonomous perception and dynamic learning of unknown vehicle models, solving the problem of manual configuration dependence in the existing technology and improving vehicle model adaptation efficiency and process execution flexibility.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- CHANGCHUN PENGTENG TECH CO LTD
- Filing Date
- 2026-03-09
- Publication Date
- 2026-06-02
AI Technical Summary
Existing technology cannot enable brake fluid filling equipment to autonomously sense and dynamically learn unknown vehicle models, resulting in the need for manual configuration for each vehicle model adaptation. It also cannot cope with protocol fine-tuning for different batches of firmware upgrades, resulting in poor adaptability.
By generating multi-dimensional protocol feature fingerprints, using a pre-trained feature association network to deconstruct and identify key protocol feature vectors, generating protocol instruction templates, and combining real-time physical parameters to dynamically instantiate communication instruction frames, autonomous perception and dynamic adjustment of unknown vehicle models can be achieved.
This technology enables the brake fluid filling equipment to communicate autonomously with unknown vehicle models, shortening the vehicle model adaptation cycle, reducing manual intervention costs, and improving the flexibility of the production line and the flexibility of process execution.
Smart Images
Figure CN122131747A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of automotive intelligent manufacturing technology, and in particular relates to a control method and system for vehicle brake fluid filling process. Background Technology
[0002] In the automotive assembly process, the vacuum filling of brake fluid is a core step in ensuring driving safety. This step requires the filling equipment to establish a real-time and reliable communication link with the vehicle's built-in Electronic Stability Program (ESP) or Anti-lock Braking System (ABS) control unit to collaboratively complete precise operations such as vacuuming, pressure maintenance, metered filling, and bleed.
[0003] Existing technologies generally employ communication modes of hard coding or parameterized querying. The refueling equipment control system pre-stores communication protocol stacks for specific vehicle models and control unit part numbers. When a new vehicle model is imported, whether the protocol is recompiled (hard coding) or a new instruction set is manually configured in the database (parameterized querying), it relies on the complete cracking and manual input of the target vehicle's communication protocol (such as CAN instruction format, diagnostic address, encryption algorithm) in the early stages.
[0004] The aforementioned existing technologies have the following inherent defects: Firstly, existing solutions rely on the premise that the device will completely fail for vehicles with unknown or unpreset protocols. Their adaptability depends entirely on manual protocol parsing and configuration after the fact, and they cannot autonomously perceive and learn unknown protocols during the initial connection. Secondly, at the process execution level, existing technologies still passively invoke "static instructions" as "process steps." Instruction generation is a one-time event, lacking an intelligent mechanism to dynamically adjust the instruction encoding structure or communication strategy based on real-time vehicle feedback during the refueling process. Thirdly, even with existing parameterized query technology, the communication instructions for each vehicle model exist as isolated data entries. The system cannot understand the inherent relationships between different instructions, nor can it deduce or predict the compositional rules of another unknown protocol from a known protocol. This results in each adaptation being a configuration task starting from scratch. Therefore, existing parameterized queries not only rely on manual intervention but also cannot handle protocol fine-tuning caused by firmware upgrades of different batches for the same vehicle model.
[0005] In summary, how to enable the filling equipment to autonomously perceive, dynamically learn, and intelligently weave vehicle communication protocols, fundamentally eliminating the reliance on pre-set protocol libraries and manual configuration, and thus better controlling the brake fluid filling process, has become a pressing technical challenge in this field. Summary of the Invention
[0006] The purpose of this invention is to provide a control method for the vehicle brake fluid filling process, thereby solving the above-mentioned technical problems.
[0007] This invention is implemented as follows: a control method for a vehicle brake fluid filling process, applied to brake fluid filling equipment, specifically including the following steps:
[0008] In response to the establishment of a communication link between the brake fluid filling device and the target vehicle, a protocol detection sequence is sent to the target vehicle's brake control system to obtain a multi-dimensional protocol feature fingerprint that characterizes the uniqueness of the target vehicle's communication protocol.
[0009] The multidimensional protocol feature fingerprint is input into a pre-trained feature association network for deconstruction, and key protocol feature vectors related to the brake fluid filling process node are identified. Based on the key protocol feature vectors, a protocol instruction template is generated.
[0010] Based on the preset standard refueling process, at each process node, the protocol instruction template is called, and combined with the real-time physical parameters of the current process, a communication instruction frame conforming to the current target vehicle communication protocol is dynamically instantiated and generated.
[0011] The system receives the execution status information of the target vehicle braking control system on the communication command frame, adjusts the action of the physical actuator according to the execution status information, and performs online incremental updates to the feature association network.
[0012] Furthermore, in response to the establishment of a communication link between the brake fluid filling device and the target vehicle, the step of sending a protocol detection sequence to the target vehicle's brake control system to obtain a multi-dimensional protocol feature fingerprint used to characterize the uniqueness of the target vehicle's communication protocol specifically includes:
[0013] Generate a set of protocol probe sequences consisting of different diagnostic session identifiers, service identifiers and sub-function codes, and send them one by one to the target vehicle braking control system according to a preset timing sequence;
[0014] Capture the raw timing response signal returned by the target vehicle's braking control system for each protocol detection sequence;
[0015] Feature engineering is performed on all captured raw time-series response signals to extract multidimensional protocol feature fingerprints.
[0016] Furthermore, the multidimensional protocol feature fingerprint includes:
[0017] Timing response characteristics: the response delay time series of each protocol probe sequence;
[0018] Padding bit encoding characteristics: The data link layer padding bit sequence is a fixed-length vector formed by hashing or statistical encoding;
[0019] Data content entropy characteristics: Entropy values and byte distribution statistics of application layer data content.
[0020] Session interaction pattern characteristics: number of interactions and timeouts in different diagnostic sessions.
[0021] Further, the step of inputting the multi-dimensional protocol feature fingerprint into a pre-trained feature association network for deconstruction, identifying key protocol feature vectors related to the brake fluid filling process node, and generating a protocol instruction template based on the key protocol feature vectors specifically includes:
[0022] The multidimensional protocol feature fingerprint is input into a pre-trained feature association network; the feature association network first maps the multidimensional protocol feature fingerprint into feature points in the latent space through an encoder;
[0023] In the latent space, key protocol feature vectors related to the brake fluid filling process node are identified based on an attention mechanism, and redundant features unrelated to the process node are extracted.
[0024] The key protocol feature vector is input into the decoder to reconstruct and generate a protocol instruction template that matches the target vehicle and contains complete handshake and execution logic.
[0025] Furthermore, the protocol instruction template is described in an abstract instruction frame format with wildcards, and includes instruction transmission timing, frame interval, error retry logic, verification parameters, and instruction format set for process nodes.
[0026] Furthermore, based on the preset standard refueling process flow, at each process node, the protocol instruction template is invoked, and combined with the real-time physical parameters of the current process, a communication instruction frame conforming to the current target vehicle communication protocol is dynamically instantiated. This specifically includes:
[0027] A process request is generated based on the current process node; the process request includes a node identifier and the real-time physical parameters of the current process node;
[0028] Based on the process request, the corresponding abstract instruction frame is retrieved from the protocol instruction template according to the current node identifier, and the real-time physical parameters are filled into the wildcard position of the abstract instruction frame to complete the first instantiation of the abstract instruction frame.
[0029] The abstract instruction frame after initial instantiation is dynamically verified, and the calculated verification value is filled into the corresponding field of the abstract instruction frame to generate a communication instruction frame that conforms to the current target vehicle communication protocol.
[0030] Another object of the present invention is to provide a control system for a vehicle brake fluid filling process, for implementing the above-mentioned control method for the vehicle brake fluid filling process, specifically including:
[0031] The protocol detection module is used to send a protocol detection sequence to the target vehicle's braking control system in response to the establishment of a communication link between the brake fluid filling device and the target vehicle, and to obtain a multi-dimensional protocol feature fingerprint to characterize the uniqueness of the target vehicle's communication protocol.
[0032] The protocol instruction template generation module is used to input the multi-dimensional protocol feature fingerprint into a pre-trained feature association network for deconstruction, identify the key protocol feature vectors related to the brake fluid filling process node, and generate a protocol instruction template based on the key protocol feature vectors.
[0033] The communication instruction frame generation module is used to dynamically instantiate and generate a communication instruction frame that conforms to the current target vehicle communication protocol at each process node according to the preset standard refueling process flow and in combination with the real-time physical parameters of the current process.
[0034] The feedback update module is used to receive the execution status information of the target vehicle braking control system on the communication command frame, adjust the action of the physical actuator according to the execution status information, and perform online incremental updates to the feature association network.
[0035] Furthermore, the protocol detection module specifically includes:
[0036] The protocol probe sequence generation unit is used to generate a set of protocol probe sequences consisting of different diagnostic session identifiers, service identifiers and sub-function codes, and send them to the target vehicle braking control system one by one according to a preset timing sequence.
[0037] The response signal acquisition unit is used to capture the raw timing response signal returned by the target vehicle braking control system for each protocol detection sequence;
[0038] The feature extraction unit is used to perform feature engineering processing on all captured raw time-series response signals to extract multi-dimensional protocol feature fingerprints.
[0039] Furthermore, the protocol instruction template generation module specifically includes:
[0040] A feature input unit is used to input the multidimensional protocol feature fingerprint into a pre-trained feature association network; the feature association network first maps the multidimensional protocol feature fingerprint into feature points in the latent space through an encoder.
[0041] The feature recognition unit is used to identify key protocol feature vectors related to the brake fluid filling process node in the latent space based on an attention mechanism, and to remove redundant features that are irrelevant to the process node.
[0042] The feature reconstruction unit is used to input the key protocol feature vector into the decoder and reconstruct a protocol instruction template that matches the target vehicle and contains complete handshake and execution logic.
[0043] Furthermore, the communication instruction frame generation module specifically includes:
[0044] A process request generation unit is used to generate a process request based on the current process node; the process request includes a node identifier and the real-time physical parameters of the current process node.
[0045] The instruction instantiation unit is used to retrieve the corresponding abstract instruction frame from the protocol instruction template based on the process request and the current node identifier, and fill the real-time physical parameters into the wildcard position of the abstract instruction frame to complete the first instantiation of the abstract instruction frame.
[0046] The instruction verification unit is used to perform dynamic verification calculations on the abstract instruction frame after its initial instantiation, and fill the calculated verification value into the corresponding fields of the abstract instruction frame to generate a communication instruction frame that conforms to the current target vehicle communication protocol.
[0047] This invention provides a control method for a vehicle brake fluid filling process. By extracting multi-dimensional protocol feature fingerprints, it achieves autonomous perception of communication protocols for unknown vehicle models. Through the deconstruction and reconstruction of feature association networks, it completes a nonlinear mapping from low-level features to high-level instruction templates. In addition, this invention achieves real-time coordination between process logic and communication protocols through a dynamic instantiation mechanism, enabling the brake fluid filling equipment to autonomously explore and successfully establish communication when facing new or variant vehicle models. This invention not only significantly shortens the vehicle model adaptation cycle and reduces manual intervention costs, but also gives the production line the flexibility to cope with unknown changes in future vehicle model protocols. Attached Figure Description
[0048] Figure 1 This is a flowchart illustrating the control method for the vehicle brake fluid filling process provided in an embodiment of the present invention.
[0049] Figure 2 This is a flowchart illustrating step S100 in the control method for vehicle brake fluid filling process provided in an embodiment of the present invention.
[0050] Figure 3 This is a flowchart illustrating step S200 in the control method for vehicle brake fluid filling process provided in an embodiment of the present invention.
[0051] Figure 4 This is a flowchart illustrating step S300 in the control method for vehicle brake fluid filling process provided in an embodiment of the present invention.
[0052] Figure 5This is a schematic diagram of the feature association network provided in an embodiment of the present invention.
[0053] Figure 6 This is a schematic diagram of the control system for the vehicle brake fluid filling process provided in an embodiment of the present invention.
[0054] Figure 7 This is a schematic diagram of the protocol detection module provided in an embodiment of the present invention.
[0055] Figure 8 This is a schematic diagram of the protocol instruction template generation module provided in an embodiment of the present invention.
[0056] Figure 9 This is a schematic diagram of the communication instruction frame generation module provided in an embodiment of the present invention. Detailed Implementation
[0057] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.
[0058] like Figure 1 As shown, in one embodiment of the present invention, a method for controlling a vehicle brake fluid filling process is provided, applied to a brake fluid filling equipment, specifically including the following steps:
[0059] S100: In response to the establishment of a communication link between the brake fluid filling device and the target vehicle, a protocol detection sequence is sent to the target vehicle's brake control system to obtain a multi-dimensional protocol feature fingerprint used to characterize the uniqueness of the target vehicle's communication protocol.
[0060] S200. Input the multi-dimensional protocol feature fingerprint into a pre-trained feature association network for deconstruction, identify the key protocol feature vectors related to the brake fluid filling process node, and generate a protocol instruction template based on the key protocol feature vectors.
[0061] S300. Based on the preset standard refueling process, at each process node, the protocol instruction template is called, and combined with the real-time physical parameters of the current process, a communication instruction frame conforming to the current target vehicle communication protocol is dynamically instantiated and generated.
[0062] S400: Receive the execution status information of the target vehicle braking control system on the communication command frame, adjust the action of the physical actuator according to the execution status information, and perform online incremental updates to the feature association network.
[0063] like Figure 2As shown, in a preferred embodiment of the present invention, in response to the establishment of a communication link between the brake fluid filling device and the target vehicle, a protocol detection sequence is sent to the target vehicle's brake control system to obtain a multi-dimensional protocol feature fingerprint for characterizing the uniqueness of the target vehicle's communication protocol. Specifically, step S100 includes:
[0064] S110. Generate a set of protocol probe sequences consisting of different diagnostic session identifiers, service identifiers and sub-function codes, and send them one by one to the target vehicle braking control system according to the preset timing sequence.
[0065] S120. Capture the raw timing response signal returned by the target vehicle's braking control system for each protocol detection sequence;
[0066] S130. Perform feature engineering on all captured raw timing response signals to extract multidimensional protocol feature fingerprints.
[0067] It should be noted that the communication link can be an OBD communication link; the protocol probe sequence refers to a set of instructions used to trigger different communication behaviors of the vehicle, which aims to obtain the protocol characteristics of the vehicle from multiple dimensions; the original timing response signal includes at least the response timestamp, the response data frame length, the data link layer padding bit sequence, and the application layer data content.
[0068] In a practical application, a car assembly line wants to perform brake fluid filling on a brand-new model of a vehicle, whose brake control system is of a specific model. The brake fluid filling equipment has never worked with this vehicle model before and has no pre-set matching communication protocol information. First, multiple protocol probe sequences are sent to the target vehicle's brake control system to perform the following probing actions: attempting to establish a default diagnostic session, attempting to read the ECU identifier, attempting a secure access request, and attempting to probe different CAN identifiers, etc.; each protocol probe sequence is sent sequentially according to a preset timing sequence (e.g., 10ms interval). Next, the raw timing response signals returned by the vehicle for some of the protocol probe sequences are received, including the response timestamp (i.e., the response arrival time, such as a delay of 5ms or 12ms from sending to receiving), the response data frame length (e.g., 8 bytes or 16 bytes), the data link layer padding bit sequence (e.g., some protocols insert padding bits "010" after specific bytes), and the application layer data content (e.g., the returned byte sequence). Then, through statistical and feature engineering processing of all raw time-series response signals, a set of numerical feature vectors is generated, ultimately forming the multidimensional protocol feature fingerprint of the vehicle; wherein, the multidimensional protocol feature fingerprint is the feature vector subsequently passed to the feature association network, specifically including:
[0069] Timing response characteristics: the response delay time series of each protocol probe sequence;
[0070] Padding bit encoding characteristics: A fixed-length vector formed by hashing or statistical encoding of the data link layer padding bit sequence (i.e., the binary sequence of padding bits in the data link layer).
[0071] Data content entropy characteristics: Entropy values and byte distribution statistics of application layer data content.
[0072] Session interaction pattern characteristics: Session-level characteristics such as the number of interactions and timeouts in different diagnostic sessions.
[0073] In this embodiment of the invention, by actively sending a protocol detection sequence to the target vehicle's braking control system and capturing the original timing response signal, the underlying characteristics of the target vehicle's communication protocol are directly perceived. This does not require relying on preset part numbers or known protocol libraries. Even when faced with completely unknown vehicle models, it can obtain their unique communication identity from the signal level, laying a data foundation for subsequent protocol deconstruction and reconstruction, and getting rid of the passive dependence of traditional technologies on preset information.
[0074] like Figure 3 As shown, in a preferred embodiment of the present invention, the step of inputting the multi-dimensional protocol feature fingerprint into a pre-trained feature association network for deconstruction, identifying key protocol feature vectors related to the brake fluid filling process node, and generating a protocol instruction template based on the key protocol feature vectors, specifically step S200, includes:
[0075] S210. Input the multidimensional protocol feature fingerprint into a pre-trained feature association network; the feature association network first maps the multidimensional protocol feature fingerprint into feature points in the latent space through an encoder;
[0076] S220. In the hidden space, based on the attention mechanism, key protocol feature vectors related to the brake fluid filling process nodes (including initialization, vacuuming, pressure holding, filling, and back suction) are identified, and redundant features unrelated to the process node are extracted.
[0077] S230. Input the key protocol feature vector into the decoder to reconstruct and generate a protocol instruction template that matches the target vehicle and contains complete handshake and execution logic; wherein, the protocol instruction template is an abstract rule, specifically described in an abstract instruction frame format with wildcards, and includes instruction sending timing, frame interval, error retry logic, verification parameters, and instruction format set of process nodes.
[0078] In practical applications, such as Figure 5 As shown, the feature association network adopts an encoder-decoder framework, and a cross-modal attention module is introduced between the encoder and decoder to enhance the extraction and reconstruction of key protocol features. The specific structure of the feature association network includes:
[0079] 1. Input layer: used to receive multidimensional protocol feature fingerprints; the multidimensional protocol features are normalized and then concatenated into a high-dimensional input vector, which is used as the input to the encoder.
[0080] II. Encoder: Consists of multiple stacked Transformer encoding layers, designed to extract deep semantic representations from input features. Each layer contains two sub-layers:
[0081] Multi-head self-attention mechanism: used to capture the dependencies between different dimensions within a feature. Since different dimensions of protocol features (such as delay and padding bits) may have implicit relationships, self-attention can model such global dependencies;
[0082] Feedforward neural networks: used to perform nonlinear transformations on the output of multi-head self-attention mechanisms to enhance the expressive power of the model;
[0083] The encoder ultimately outputs a latent space representation that condenses the core information of the input multidimensional protocol feature fingerprint and removes noise and redundancy.
[0084] III. Cross-modal Attention Module: Located between the encoder and decoder, this module identifies key protocol feature vectors strongly correlated with the brake fluid filling process nodes. The cross-modal attention module uses a latent space representation as the query and predefined process node embedding vectors (e.g., mapping node types to vectors through word embedding layers) as keys and values. Through attention calculation, it obtains a weighted feature vector, which is the key protocol feature vector. This mechanism ensures that the decoder focuses on protocol features relevant to the current process when generating protocol instruction templates, ignoring irrelevant information.
[0085] IV. Decoder: Composed of multiple Transformer decoding layers, its goal is to progressively decode key protocol feature vectors into an abstract protocol instruction template. The decoder's output sequence is an abstract instruction frame, containing fixed fields (such as the protocol header) and variable fields (represented by wildcards); simultaneously, the decoder also outputs metadata associated with this protocol instruction template, including:
[0086] Command transmission timing: the time interval between adjacent commands;
[0087] Frame interval: The interval between transmissions of bytes within the same instruction;
[0088] Error retry logic: the number of retries and the backoff time when there is no response or the response is incorrect;
[0089] Verification parameters: Define the verification rules applicable to the target vehicle communication protocol, including at least the verification algorithm type, initial value, verification value length, and specified insertion or appending position;
[0090] The instruction format set for process nodes: records the abstract instruction skeleton corresponding to each brake fluid filling process node (such as initialization, vacuuming, pressure holding, filling, and back suction); the abstract instruction skeleton contains fixed fields and variable parameter placeholders.
[0091] V. Output Layer: This layer maps the sequences generated by the decoder to specific protocol instruction template formats, along with the aforementioned metadata, and outputs the final protocol instruction template.
[0092] Furthermore, the training process of the feature association network is based on a large number of multi-dimensional protocol feature fingerprints of known vehicle models and their corresponding successful protocol instruction sets as training pairs. Through unsupervised learning, a non-linear mapping relationship between features and instructions is constructed. Specifically, the feature association network adopts a two-stage training strategy: first, large-scale unsupervised pre-training, followed by small-sample supervised fine-tuning, combined with online incremental updates, to ensure the network's generalization ability and adaptability. This includes the following steps:
[0093] (1) Construction of the training dataset: The training data comes from a large number of known vehicle models' refueling interaction records accumulated during historical production. Each record contains a multi-dimensional protocol feature fingerprint and a successful protocol instruction set (a sequence of communication instruction frames that have been actually used and verified to be successful in the subsequent refueling process of that vehicle model). The multi-dimensional protocol feature fingerprint in each record and the corresponding successful protocol instruction set are combined to form a training pair to obtain the training dataset.
[0094] (2) Unsupervised pre-training (first stage): In the unsupervised pre-training stage, the idea of masked autoencoder is used to train the encoder-decoder. Specifically, the encoder is required to reconstruct the complete feature fingerprint (not the instruction template) by randomly masking a portion of the input multidimensional protocol feature fingerprint. This task forces the encoder to learn the statistical structure and implicit rules inside the features, thereby forming a good latent space representation; after pre-training, the encoder has the ability to extract the deep semantics of the protocol features.
[0095] (3) Supervised fine-tuning (second stage): Based on the above pre-training, the above training pairs are loaded for supervised fine-tuning; the fine-tuning objective is to minimize the weighted sum of the following loss functions:
[0096] Command frame reconstruction loss: Calculate the cross-entropy loss (for discrete protocol bytes) or mean square error loss (for consecutive wildcard positions) between the protocol command template output by the decoder and the real template.
[0097] Time series prediction loss: the mean square error between the predicted instruction interval and the actual interval;
[0098] Error retry logic loss: classification loss (class prediction of the retry strategy).
[0099] Then, the network parameters are optimized through backpropagation, enabling the network to accurately map from multi-dimensional protocol feature fingerprints to corresponding protocol instruction templates.
[0100] (4) Nonlinear mapping mechanism from features to instructions: The feature association network learns a nonlinear mapping relationship, which is to deconstruct and reorganize features through the latent space. Since the protocols of different models of the same brand may have similarities, the network can use this similarity to transfer knowledge: even if a certain model has never been seen, as long as its multidimensional protocol feature fingerprint is close to that of some known models in the latent space, the network can generate a reasonable protocol instruction template and fine-tune it through feedback in subsequent interactions.
[0101] (5) Online Incremental Update (Continuous Learning): In practical applications, the feature association network is not static; each time an annotation interaction is successful, the system will use the multidimensional protocol feature fingerprint and the corresponding successful protocol instruction set as new samples to perform online incremental updates to the feature association network; the update methods include, but are not limited to:
[0102] Mini-batch gradient descent: Optimizes network parameters using a small number of iterations with new samples;
[0103] Experience review: Combining important historical examples to prevent catastrophic amnesia;
[0104] Reinforcement learning assistance: When communication fails, the protocol probe sequence is adjusted using a reinforcement learning strategy, and the successful path is used as a positive sample to update the feature association network;
[0105] Through this continuous learning mechanism, the feature association network can constantly adapt to changes brought about by new protocols, protocol variants, and vehicle firmware upgrades.
[0106] It should be noted that, in addition to the framework described above, feature association networks can also employ convolutional neural networks, recurrent neural networks, or other machine learning models capable of mapping features to instructions; they are not limited to these.
[0107] In this embodiment of the invention, the key protocol feature vectors related to the brake fluid filling process can be identified through the feature association network, and redundant information can be automatically removed. An abstract protocol instruction template can be generated, thereby realizing intelligent translation from the underlying signal features to the high-level communication protocol. This enables the system to understand and predict the composition rules of unknown protocols, breaking through the limitations of isolated instruction data and inability to generalize in traditional parameterized configuration.
[0108] like Figure 4As shown, in a preferred embodiment of the present invention, according to a preset standard refueling process, at each process node, the protocol instruction template is invoked, and a communication instruction frame conforming to the current target vehicle communication protocol is dynamically instantiated and generated in combination with the real-time physical parameters of the current process. Specifically, step S300 includes:
[0109] S310. Generate a process request based on the current process node; the process request includes a node identifier and real-time physical parameters of the current process node (such as target vacuum level, holding time, etc.).
[0110] S320. Based on the process request, retrieve the corresponding abstract instruction frame from the protocol instruction template according to the current node identifier, and fill the real-time physical parameters into the wildcard position of the abstract instruction frame to complete the first instantiation of the abstract instruction frame.
[0111] S330. Perform dynamic verification calculation on the abstract instruction frame after initial instantiation, and fill the calculated verification value into the corresponding field of the abstract instruction frame to generate a communication instruction frame that conforms to the current target vehicle communication protocol; at the same time, send the communication instruction frame to the target vehicle braking control system through the OBD interface according to the instruction sending timing defined by the protocol instruction template.
[0112] Specifically, the communication command frame is constructed according to the verification rules of the target vehicle communication protocol in the verification parameters (such as verification algorithm type, initial value, verification value length, and specified insertion or appending position); the dynamic verification calculation method can adopt verification algorithms such as cumulative sum verification, XOR verification, or cyclic redundancy check, which can be determined according to the verification parameters in the protocol command template; this embodiment of the invention takes the cumulative sum verification method as an example, assuming that the abstract command frame after the initial instantiation is D=[d0,d1,…,d L−1 ], where d i This represents the i-th byte (0≤d) in the abstract instruction frame after initial instantiation. i ≤255), where L is the number of bytes in the abstract instruction frame; the formula for calculating the checksum C is:
[0113] ;
[0114] In the formula, `init` is the initial accumulation value, usually 0x00 or 0xFF, which can be determined according to the check parameters in the protocol instruction template; `cs_len` is the byte length of the check value, usually 1 or 2, which can be determined according to the check parameters in the protocol instruction template; and `mod` is the modulo operation. Through the above verification, the check value is converted into a byte sequence of length `cs_len`, and inserted or appended to the specified position of the abstract instruction frame according to the byte order required by the protocol (big-endian or little-endian, where big-endian means the most significant byte comes first, and little-endian means the least significant byte comes first), forming a precise communication instruction frame.
[0115] In practical applications, suppose the protocol requirements of a certain vehicle model are as follows: the abstract instruction frame after initial instantiation is D=[0x01,0x02,0x03]; init=0x00; cs_len=1; the byte order is big-endian; the specified position for insertion or appending is the end; then according to the above cumulative check method, the check value C=0x06 can be obtained. Appending this check value to the end of the abstract instruction frame will yield the accurate communication instruction frame F=[0x01,0x02,0x03,0x06].
[0116] It is worth noting that during the communication process between the brake fluid filling device and the vehicle's brake control system, the check value is an additional data piece attached to the end of the communication command frame or other designated locations. Its core function is to ensure the integrity, accuracy, and reliability of the communication data.
[0117] In this embodiment of the invention, by transforming the abstract instruction frame into a specific communication instruction frame that conforms to the target vehicle specification, the decoupling of process logic and communication protocol is achieved. At the same time, the real-time accuracy and context adaptability of the instructions are ensured, enabling the brake fluid filling equipment to accurately coordinate with the vehicle control system to complete complex filling procedures, significantly improving the flexibility and reliability of process execution.
[0118] In a preferred embodiment of the present invention, the execution status information includes confirmation information of successful / failed instruction execution and possible status data. The status data in the execution status information is forwarded to the physical actuators of the brake fluid filling equipment (such as vacuum pumps, filling valves, NC valves, etc.) and parsed to determine whether to continue to the next process or to handle the abnormality until the process flow ends. For example, in practical applications, the power of the vacuum pump is adjusted in real time according to the process node (vacuuming) in the status data and the real-time physical parameters (target vacuum degree) of the current process node to complete the vacuuming process of brake fluid filling. Then, if the vacuuming is confirmed to be successful, the brake fluid filling equipment is controlled to continue to execute the next pressure holding process. Simultaneously, the execution status information and the communication command frame used in this communication are sent to the control system of the brake fluid filling device. The communication command frame is associated with the previously extracted multi-dimensional protocol feature fingerprint to form positive samples (if communication is successful) or negative samples (if communication fails), thus obtaining a sample set. The above feature association network is updated online using this sample set to optimize the subsequent deconstruction and reconstruction accuracy of the protocol for the same type of vehicle. When communication fails, a reinforcement learning strategy can be initiated to adjust the protocol detection sequence in the above steps and re-execute the above steps until communication is successful.
[0119] like Figure 6 As shown, in another embodiment of the present invention, a control system for a vehicle brake fluid filling process is also provided, which implements the above-described control method for the vehicle brake fluid filling process, specifically including:
[0120] Protocol detection module 10 is used to send a protocol detection sequence to the target vehicle's braking control system in response to the establishment of a communication link between the brake fluid filling device and the target vehicle, and to obtain a multi-dimensional protocol feature fingerprint to characterize the uniqueness of the target vehicle's communication protocol.
[0121] The protocol instruction template generation module 20 is used to input the multi-dimensional protocol feature fingerprint into a pre-trained feature association network for deconstruction, identify the key protocol feature vectors related to the brake fluid filling process node, and generate a protocol instruction template based on the key protocol feature vectors.
[0122] The communication instruction frame generation module 30 is used to call the protocol instruction template at each process node according to the preset standard filling process flow, and dynamically instantiate and generate a communication instruction frame that conforms to the current target vehicle communication protocol in combination with the real-time physical parameters of the current process.
[0123] The feedback update module 40 is used to receive the execution status information of the target vehicle braking control system on the communication command frame, adjust the action of the physical actuator according to the execution status information, and perform online incremental updates to the feature association network.
[0124] like Figure 7 As shown, in a preferred embodiment of the present invention, the protocol detection module 10 specifically includes:
[0125] The protocol probe sequence generation unit 11 is used to generate a set of protocol probe sequences consisting of different diagnostic session identifiers, service identifiers and sub-function codes, and send them to the target vehicle braking control system one by one according to a preset timing sequence.
[0126] The response signal acquisition unit 12 is used to acquire the original timing response signal returned by the target vehicle braking control system for each protocol detection sequence;
[0127] The feature extraction unit 13 is used to perform feature engineering processing on all captured raw time-series response signals to extract multidimensional protocol feature fingerprints.
[0128] like Figure 8 As shown, in a preferred embodiment of the present invention, the protocol instruction template generation module 20 specifically includes:
[0129] Feature input unit 21 is used to input the multidimensional protocol feature fingerprint into a pre-trained feature association network; the feature association network first maps the multidimensional protocol feature fingerprint into feature points in the latent space through an encoder;
[0130] The feature recognition unit 22 is used to identify key protocol feature vectors related to the brake fluid filling process node in the latent space based on an attention mechanism, and to remove redundant features that are unrelated to the process node.
[0131] The feature reconstruction unit 23 is used to input the key protocol feature vector into the decoder and reconstruct a protocol instruction template that matches the target vehicle and contains complete handshake and execution logic.
[0132] like Figure 9 As shown, in a preferred embodiment of the present invention, the communication command frame generation module 30 specifically includes:
[0133] The process request generation unit 31 is used to generate a process request based on the current process node; the process request includes a node identifier and the real-time physical parameters of the current process node.
[0134] The instruction instantiation unit 32 is used to retrieve the corresponding abstract instruction frame from the protocol instruction template based on the process request and the current node identifier, and fill the real-time physical parameters into the wildcard position of the abstract instruction frame to complete the first instantiation of the abstract instruction frame.
[0135] The instruction verification unit 33 is used to perform dynamic verification calculations on the abstract instruction frame after its initial instantiation, and fill the calculated verification value into the corresponding field of the abstract instruction frame to generate a communication instruction frame that conforms to the current target vehicle communication protocol.
[0136] It should be noted that the above modules and units can be implemented as a computer program, which can run on a computer device. The computer device's memory can store the computer program that makes up the modules or units, enabling the processor to execute the various steps of the above method.
[0137] It should be understood that although the steps in the flowcharts of the embodiments of the present invention are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in each embodiment may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least a portion of the sub-steps or stages of other steps.
[0138] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods.
[0139] The above embodiments merely illustrate several implementation methods of the present invention, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of this patent should be determined by the appended claims.
Claims
1. A control method for a vehicle brake fluid filling process, applied to brake fluid filling equipment, characterized in that, Includes the following steps: In response to the establishment of a communication link between the brake fluid filling device and the target vehicle, a protocol detection sequence is sent to the target vehicle's brake control system to obtain a multi-dimensional protocol feature fingerprint that characterizes the uniqueness of the target vehicle's communication protocol. The multidimensional protocol feature fingerprint is input into a pre-trained feature association network for deconstruction, and key protocol feature vectors related to the brake fluid filling process node are identified. Based on the key protocol feature vectors, a protocol instruction template is generated. Based on the preset standard refueling process, at each process node, the protocol instruction template is called, and combined with the real-time physical parameters of the current process, a communication instruction frame conforming to the current target vehicle communication protocol is dynamically instantiated and generated. The system receives the execution status information of the target vehicle braking control system on the communication command frame, adjusts the action of the physical actuator according to the execution status information, and performs online incremental updates to the feature association network.
2. The control method for the vehicle brake fluid filling process according to claim 1, characterized in that, The steps of responding to the establishment of a communication link between the brake fluid filling device and the target vehicle, sending a protocol probe sequence to the target vehicle's brake control system, and obtaining a multi-dimensional protocol feature fingerprint to characterize the uniqueness of the target vehicle's communication protocol specifically include: Generate a set of protocol probe sequences consisting of different diagnostic session identifiers, service identifiers and sub-function codes, and send them one by one to the target vehicle braking control system according to a preset timing sequence; Capture the raw timing response signal returned by the target vehicle's braking control system for each protocol detection sequence; Feature engineering is performed on all captured raw time-series response signals to extract multidimensional protocol feature fingerprints.
3. The control method for the vehicle brake fluid filling process according to claim 2, characterized in that, The multidimensional protocol feature fingerprint includes: Timing response characteristics: the response delay time series of each protocol probe sequence; Padding bit encoding characteristics: The data link layer padding bit sequence is a fixed-length vector formed by hashing or statistical encoding; Data content entropy characteristics: Entropy values and byte distribution statistics of application layer data content. Session interaction pattern characteristics: number of interactions and timeouts in different diagnostic sessions.
4. The control method for the vehicle brake fluid filling process according to claim 1, characterized in that, The steps of inputting the multidimensional protocol feature fingerprint into a pre-trained feature association network for deconstruction, identifying key protocol feature vectors related to the brake fluid filling process node, and generating a protocol instruction template based on the key protocol feature vectors specifically include: The multidimensional protocol feature fingerprint is input into a pre-trained feature association network; the feature association network first maps the multidimensional protocol feature fingerprint into feature points in the latent space through an encoder; In the latent space, key protocol feature vectors related to the brake fluid filling process node are identified based on an attention mechanism, and redundant features unrelated to the process node are extracted. The key protocol feature vector is input into the decoder to reconstruct and generate a protocol instruction template that matches the target vehicle and contains complete handshake and execution logic.
5. The control method for the vehicle brake fluid filling process according to claim 4, characterized in that, The protocol instruction template is described in an abstract instruction frame format with wildcards, and includes instruction transmission timing, frame interval, error retry logic, verification parameters, and instruction format set for process nodes.
6. The control method for the vehicle brake fluid filling process according to claim 5, characterized in that, According to the preset standard refueling process, at each process node, the protocol instruction template is invoked, and a communication instruction frame conforming to the current target vehicle communication protocol is dynamically instantiated and generated in combination with the real-time physical parameters of the current process. Specifically, this includes: A process request is generated based on the current process node; the process request includes a node identifier and the real-time physical parameters of the current process node; Based on the process request, the corresponding abstract instruction frame is retrieved from the protocol instruction template according to the current node identifier, and the real-time physical parameters are filled into the wildcard position of the abstract instruction frame to complete the first instantiation of the abstract instruction frame. The abstract instruction frame after initial instantiation is dynamically verified, and the calculated verification value is filled into the corresponding field of the abstract instruction frame to generate a communication instruction frame that conforms to the current target vehicle communication protocol.
7. A control system for a vehicle brake fluid filling process, used to implement the control method for the vehicle brake fluid filling process according to any one of claims 1-6, characterized in that, include: The protocol detection module is used to send a protocol detection sequence to the target vehicle's braking control system in response to the establishment of a communication link between the brake fluid filling device and the target vehicle, and to obtain a multi-dimensional protocol feature fingerprint to characterize the uniqueness of the target vehicle's communication protocol. The protocol instruction template generation module is used to input the multi-dimensional protocol feature fingerprint into a pre-trained feature association network for deconstruction, identify the key protocol feature vectors related to the brake fluid filling process node, and generate a protocol instruction template based on the key protocol feature vectors. The communication instruction frame generation module is used to dynamically instantiate and generate a communication instruction frame that conforms to the current target vehicle communication protocol at each process node according to the preset standard refueling process flow and in combination with the real-time physical parameters of the current process. The feedback update module is used to receive the execution status information of the target vehicle braking control system on the communication command frame, adjust the action of the physical actuator according to the execution status information, and perform online incremental updates to the feature association network.
8. The control system for the vehicle brake fluid filling process according to claim 7, characterized in that, The protocol detection module specifically includes: The protocol probe sequence generation unit is used to generate a set of protocol probe sequences consisting of different diagnostic session identifiers, service identifiers and sub-function codes, and send them to the target vehicle braking control system one by one according to a preset timing sequence. The response signal acquisition unit is used to capture the raw timing response signal returned by the target vehicle braking control system for each protocol detection sequence; The feature extraction unit is used to perform feature engineering processing on all captured raw time-series response signals to extract multi-dimensional protocol feature fingerprints.
9. The control system for the vehicle brake fluid filling process according to claim 7, characterized in that, The protocol instruction template generation module specifically includes: A feature input unit is used to input the multidimensional protocol feature fingerprint into a pre-trained feature association network; the feature association network first maps the multidimensional protocol feature fingerprint into feature points in the latent space through an encoder. The feature recognition unit is used to identify key protocol feature vectors related to the brake fluid filling process node in the latent space based on an attention mechanism, and to remove redundant features that are irrelevant to the process node. The feature reconstruction unit is used to input the key protocol feature vector into the decoder and reconstruct a protocol instruction template that matches the target vehicle and contains complete handshake and execution logic.
10. The control system for the vehicle brake fluid filling process according to claim 7, characterized in that, The communication command frame generation module specifically includes: A process request generation unit is used to generate a process request based on the current process node; the process request includes a node identifier and the real-time physical parameters of the current process node. The instruction instantiation unit is used to retrieve the corresponding abstract instruction frame from the protocol instruction template based on the process request and the current node identifier, and fill the real-time physical parameters into the wildcard position of the abstract instruction frame to complete the first instantiation of the abstract instruction frame. The instruction verification unit is used to perform dynamic verification calculations on the abstract instruction frame after its initial instantiation, and fill the calculated verification value into the corresponding fields of the abstract instruction frame to generate a communication instruction frame that conforms to the current target vehicle communication protocol.