Cutter lifting height control system of electric spark machine tool
Through artificial intelligence predictive control and energy efficiency management modules, combined with deep learning models and high-precision sensors, the problems of precision and energy consumption in long-term processing of traditional electric spark machine tool tool tool lift height control system are solved, achieving efficient and accurate processing and Energy efficiency optimization.
Patent Information
- Application Number
- CN202411224410.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-03
- Publication Date
- 2024-12-06
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The tool lift height control system of traditional electric spark machine tools lacks adaptability and is difficult to cope with long-term processing and high-precision requirements, resulting in low processing efficiency, increased energy consumption and unstable product accuracy.
Using artificial intelligence predictive control module combined with deep learning model, combined with energy efficiency management module and high-precision sensor fusion system, predict future processing status and tool lifting needs through real-time data acquisition and analysis, adjust the tool lifting height in advance, and dynamically through energy efficiency optimization algorithm Adjust paths and heights to minimize energy consumption.
High-precision and high-efficiency processing is achieved, error accumulation is avoided, energy consumption is significantly reduced, equipment service life is extended, and control accuracy and energy efficiency optimization capabilities are improved through continuous learning and adjustment.
Smart Images

Figure CN119077071A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of electric spark machine tools, and in particular to a tool lifting height control system for electric spark machine tools. Background Art
[0002] Electric discharge machine (EDM) is a key equipment commonly used for processing complex shapes, hard materials and high-precision parts. Traditional EDM machines adjust the tool lift height through mechanical feedback and simple closed-loop control systems to ensure that the spark gap between the electrode and the workpiece remains within an appropriate range during processing. However, these traditional methods lack the ability to adapt to complex processing environments, making it difficult to cope with long-term processing and high-precision requirements, often resulting in low processing efficiency, increased energy consumption and unstable product accuracy.
[0003] In recent years, the rapid development of artificial intelligence and deep learning technology has provided new optimization directions for the control system of EDM machines. Through artificial intelligence prediction models and multi-sensor fusion technology, the tool lift height can be controlled more accurately to improve the stability and accuracy of processing. At the same time, introducing energy efficiency management into the control system can not only optimize the processing path, but also significantly reduce energy consumption and extend the service life of the equipment. Although some studies have attempted to apply artificial intelligence to the control of EDM machines, the solution of integrating multiple intelligent control modules, energy efficiency management, sensor fusion and human-computer interaction interface into a complete system is not yet mature. Summary of the invention
[0004] The purpose of this section is to summarize some aspects of embodiments of the present invention and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section and the specification abstract and the invention title of this application to avoid blurring the purpose of this section, the specification abstract and the invention title, and such simplifications or omissions cannot be used to limit the scope of the present invention.
[0005] In view of the above-mentioned problems existing in the existing tool lifting height control system of the electric spark machine tool, the present invention is proposed.
[0006] Therefore, the purpose of the present invention is to provide a tool lift height control system for an EDM machine tool, which, through predictive control, can adjust the tool lift height in advance to avoid the accumulation of errors, especially to ensure processing accuracy during long-term processing.
[0007] In order to solve the above technical problems, the present invention provides the following technical solutions: a tool lift height control system for an electric spark machine tool, comprising:
[0008] Artificial intelligence predictive control module, including:
[0009] Deep learning prediction model: trained based on historical processing data, current processing parameters and real-time sensor data; the model can predict future processing status and tool lifting requirements;
[0010] Data acquisition and analysis module: collects sensor data such as temperature, spark gap, current and processing data in real time, and pre-processes it for use in the prediction model;
[0011] Feedback control module: According to the output of the prediction model, the tool lift height is adjusted in advance to ensure processing accuracy and stability;
[0012] Energy efficiency management module, including:
[0013] Energy monitoring unit: monitors the energy consumption of the knife lifting action in real time and records the power usage;
[0014] Energy efficiency optimization algorithm: Combined with the results of artificial intelligence predictive control, it dynamically optimizes the tool lifting path and height to minimize energy consumption while maintaining high-precision processing;
[0015] Energy efficiency feedback system: Feedback energy consumption data to the prediction control module for further optimization of the model;
[0016] Human-computer interaction interface, including:
[0017] Intelligent operation interface: Provides a graphical interface that enables operators to monitor and adjust system parameters in real time. The interface displays the effects of predictive adjustments and energy efficiency management.
[0018] Data visualization module: provides dynamic change charts of energy consumption and tool lift height during machining to help operators optimize operation strategies;
[0019] High-precision sensor fusion system, including:
[0020] Laser distance sensor: used to detect spark gap and tool lift height in real time;
[0021] Multi-sensor fusion algorithm: used to fuse data from multiple sensors and provide accurate feedback to the artificial intelligence prediction module.
[0022] As a preferred solution of the EDM machine tool lift height control system of the present invention, the control method of the EDM machine tool lift height control system comprises the following steps:
[0023] Step 1: Initial data collection and model training
[0024] The system first collects historical data under different working conditions through multiple processing data, and then uses this data to train a deep learning model to predict possible changes and tool lifting needs in future processing;
[0025] Step 2: Real-time data monitoring and prediction
[0026] During the machining process, the energy monitoring unit collects the power consumption data of the EDM machine in real time, and the sensor system collects the tool lift height and spark gap data. These data are fed into the prediction model, which outputs the possible future spark gap changes and the required tool lift height.
[0027] Step 3: Predictive control of tool lift height
[0028] The predictive control module adjusts the tool lift height in advance based on the model's prediction results to cope with upcoming changes in working conditions. This advance adjustment can avoid a drop in machining accuracy due to sudden changes.
[0029] Step 4: Implementation of energy efficiency optimization strategy
[0030] Based on the output of predictive control and real-time energy monitoring results, the energy efficiency management module dynamically adjusts the tool lifting path and height to optimize energy usage;
[0031] Step 5: Feedback and Adaptive Learning
[0032] The system continuously learns from real-time processing data and feeds this data into the model for retraining, thereby improving prediction accuracy and energy efficiency optimization capabilities in subsequent processing.
[0033] As a preferred solution of the EDM machine tool lift height control system of the present invention, the deep learning prediction model structure combines the feature extraction capability of CNN and the time series prediction capability of LSTM, including:
[0034] Input layer: contains historical processing data such as spark gap, electrode wear, material properties and real-time sensor data such as temperature and current;
[0035] CNN layer: used to extract local features in the input data, especially the spatial correlation of processing parameters;
[0036] LSTM layer: captures time series information in the input data to predict future knife lifting needs;
[0037] Fully connected layer: Comprehensively process the outputs of CNN and LSTM to finally get the prediction results;
[0038] Output layer: gives the knife lifting height requirement in the future time period.
[0039] As a preferred solution of the EDM machine tool lift height control system of the present invention, wherein: the input layer input data is represented as X, which contains multiple variables, including:
[0040] Cremation gap data X1;
[0041] Electrode wear data X2;
[0042] Material property data X3;
[0043] Real-time sensor data X4;
[0044] These input data are represented as a three-dimensional tensor Where N is the number of samples, T is the time step, and D is the number of features per time step;
[0045] The convolutional neural network layer is the CNN layer, which is expressed as:
[0046] Z cnn =f(W cnn *X+b cnn )
[0047] Where: W cnn is the weight matrix of the convolution kernel; * indicates the convolution operation; b cnn is a bias term; f(·) is an activation function, such as ReLU, which can gradually extract features at different levels through multiple convolutional layers and pooling layers;
[0048] The long short-term memory network layer, or LSTM layer, consists of the following parts:
[0049] Input gate: i t =σ(W i ·[h t-1 , x t ]+b i )
[0050] Forget gate: f t =σ(W f ·[h t-1 , x t ]+b f )
[0051] Memory cell state: C t =f t ·C t-1 +i t tanh(W C ·[h t-1 , x t ]+b C )
[0052] Output gate: o t =σ(W o ·[h t-1 , x t ]+b o )
[0053] Hidden state: ht =o t tanh(C t )
[0054] Where σ(·) represents the Sigmoid activation function;
[0055] W i , W f , W C , W o is the corresponding weight matrix;
[0056] b i , b f , b C , b o is the corresponding bias term;
[0057] ·Represents bit-by-bit multiplication of elements;
[0058] The output Z of the LSTM layer lstm Will be used as input to the next fully connected layer;
[0059] The fully connected layer obtains the final prediction value:
[0060] Z fc =f(W fc ·[Z cnn , Z lstm ]+b fc )
[0061] Among them, W fc is the weight matrix of the fully connected layer;
[0062] b fc is the bias term;
[0063] f(·) is the activation function, choose ReLU or Sigmoid activation function;
[0064] The output layer gives the knife lifting height requirement in the future time period:
[0065]
[0066] Among them, W out and b out are the weights and biases of the output layer.
[0067] As a preferred solution of the EDM machine tool lift height control system of the present invention, wherein: a loss function needs to be used in the model training process, and the loss function usually uses the mean square error for regression tasks:
[0068]
[0069] Among them, yi It is the actual knife lifting height. is the predicted value of the model, and N is the number of samples.
[0070] As a preferred solution of the EDM machine tool lift height control system of the present invention, the prediction process of the model includes the following steps:
[0071] S1: Data input: input historical processing data and real-time sensor data;
[0072] S2: Feature extraction: extract local features through CNN and capture time series information through LSTM;
[0073] S3: Comprehensive processing: The fully connected layer processes the features comprehensively to obtain the prediction results;
[0074] S4: Output prediction: The output layer gives the future prediction value of the knife lifting height.
[0075] As a preferred solution of the EDM machine tool lift height control system of the present invention, the multi-sensor fusion algorithm uses the Kalman filter algorithm to integrate and optimize the data from different sensors, including two main steps:
[0076] Prediction: predict the state at the next moment based on the system's state model;
[0077] Update: Update the state estimate based on the sensor observations;
[0078] Set the system state vector x k , the observation vector is z k , the state equation and observation equation of the system are expressed as:
[0079] Equation of state:
[0080] x k =F k x k-1 +B k u k +W k
[0081] Among them: F k is the state transition matrix;
[0082] B k is the control input matrix;
[0083] u k is the control vector;
[0084] w k is the process noise, assumed to be zero-mean Gaussian white noise with covariance Q k ;
[0085] Observation equation:
[0086] z k =H k x k +v k
[0087] Where: H k is the observation matrix;
[0088] v k is the observation noise, assumed to be zero-mean Gaussian white noise with covariance R k .
[0089] As a preferred solution of the electric spark machine tool lift height control system of the present invention, the Kalman filtering step includes:
[0090] A1. Prediction steps:
[0091] Status prediction: x k|k-1 =F k x k-1|k-1 +B k u k
[0092] Covariance prediction: p k|k-1 =F k p k-1|k-1 F k +Q k
[0093] Where: x k|k-1 is the state prediction value at time k;
[0094] p k|k-1 is the forecast covariance matrix;
[0095] A2. Update steps:
[0096] Kalman gain calculation: K k =p k|k-1 H k (H k p k|k-1 H k +R k ) -1
[0097] Status update: x k|k =x k|k-1 +K k (z k -H k x k|k-1 )
[0098] Covariance update: p k|k =(IKk H k ) k|k-1
[0099] Among them, K k is the Kalman gain matrix, which determines the weighted ratio of prediction to observation;
[0100] x k|k is the state update value at time k;
[0101] p k|k is the updated covariance matrix.
[0102] As a preferred solution of the EDM machine tool lift height control system of the present invention, the steps for implementing multi-sensor fusion are as follows:
[0103] In the scenario of multi-sensor fusion, assuming that there are multiple sensors providing observation data, they are integrated into the Kalman filter framework in the following way:
[0104] Multi-sensor observation:
[0105] Assume that there are M sensors, and the observation equation of each sensor is expressed as:
[0106]
[0107] Observation Fusion:
[0108] The observation data of each sensor is fused into a total observation vector:
[0109]
[0110] Among them, w i is the weighting matrix of the ith sensor, usually based on the observation noise covariance of each sensor calculate;
[0111] Observation matrix fusion:
[0112] The observation matrix H k Weighted fusion is required:
[0113]
[0114] The observation data and observation matrix after the above fusion can be directly substituted into the update step of the Kalman filter to complete the multi-sensor data fusion.
[0115] Beneficial effects of the present invention:
[0116] 1. High precision and high efficiency: Through predictive control, the system can adjust the tool lift height in advance to avoid the accumulation of errors, especially to ensure processing accuracy during long-term processing.
[0117] 2. Energy saving and environmental protection: The energy efficiency management module optimizes the energy usage of the system, reduces unnecessary energy consumption, and extends the service life of the equipment.
[0118] 3. Intelligence and adaptability: The system continuously learns and adjusts to gradually improve its control accuracy and energy efficiency optimization capabilities and adapt to various processing conditions.
[0119] This architecture combines the advantages of artificial intelligence and energy efficiency management to provide an innovative solution for tool height control of EDM machines, which can significantly improve the performance and sustainability of the system. BRIEF DESCRIPTION OF THE DRAWINGS
[0120] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative labor. Among them:
[0121] Figure 1 It is a schematic diagram of the overall structure of the tool lifting height control system of the electric spark machine tool of the present invention. DETAILED DESCRIPTION
[0122] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.
[0123] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.
[0124] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.
[0125] Secondly, the present invention is described in detail with reference to the schematic diagram. When describing the embodiments of the present invention in detail, for the sake of convenience, the cross-sectional diagrams showing the device structure will not be partially enlarged according to the general scale, and the schematic diagrams are only examples, which should not limit the scope of protection of the present invention. In addition, in actual production, the three-dimensional dimensions of length, width and depth should be included.
[0126] Reference Figure 1 , provides an EDM machine tool lift height control system, including:
[0127] Artificial intelligence predictive control module, including:
[0128] Deep learning prediction model: trained based on historical processing data, current processing parameters and real-time sensor data; the model can predict future processing status and tool lifting requirements;
[0129] Data acquisition and analysis module: collects sensor data such as temperature, spark gap, current and processing data in real time, and pre-processes it for use in the prediction model;
[0130] Feedback control module: According to the output of the prediction model, the tool lift height is adjusted in advance to ensure processing accuracy and stability;
[0131] Energy efficiency management module, including:
[0132] Energy monitoring unit: monitors the energy consumption of the knife lifting action in real time and records the power usage;
[0133] Energy efficiency optimization algorithm: Combined with the results of artificial intelligence predictive control, it dynamically optimizes the tool lifting path and height to minimize energy consumption while maintaining high-precision processing;
[0134] Energy efficiency feedback system: Feedback energy consumption data to the prediction control module for further optimization of the model;
[0135] Human-computer interaction interface, including:
[0136] Intelligent operation interface: Provides a graphical interface that enables operators to monitor and adjust system parameters in real time. The interface displays the effects of predictive adjustments and energy efficiency management.
[0137] Data visualization module: provides dynamic change charts of energy consumption and tool lift height during machining to help operators optimize operation strategies;
[0138] High-precision sensor fusion system, including:
[0139] Laser distance sensor: used to detect spark gap and tool lift height in real time;
[0140] Multi-sensor fusion algorithm: used to fuse data from multiple sensors and provide accurate feedback to the artificial intelligence prediction module.
[0141] The control method of the electric spark machine tool tool lift height control system comprises the following steps:
[0142] Step 1: Initial data collection and model training
[0143] The system first collects historical data under different working conditions through multiple processing data, and then uses this data to train a deep learning model to predict possible changes and tool lifting needs in future processing;
[0144] Step 2: Real-time data monitoring and prediction
[0145] During the machining process, the energy monitoring unit collects the power consumption data of the EDM machine in real time, and the sensor system collects the tool lift height and spark gap data. These data are fed into the prediction model, which outputs the possible future spark gap changes and the required tool lift height.
[0146] Step 3: Predictive control of tool lift height
[0147] The predictive control module adjusts the tool lift height in advance based on the model's prediction results to cope with upcoming changes in working conditions. This advance adjustment can avoid a drop in machining accuracy due to sudden changes.
[0148] Step 4: Implementation of energy efficiency optimization strategy
[0149] Based on the output of predictive control and real-time energy monitoring results, the energy efficiency management module dynamically adjusts the tool lifting path and height to optimize energy usage;
[0150] Step 5: Feedback and Adaptive Learning
[0151] The system continuously learns from real-time processing data and feeds this data into the model for retraining, thereby improving prediction accuracy and energy efficiency optimization capabilities in subsequent processing.
[0152] Specifically, the deep learning prediction model structure combines the feature extraction capability of CNN and the time series prediction capability of LSTM, including:
[0153] Input layer: contains historical processing data such as spark gap, electrode wear, material properties and real-time sensor data such as temperature and current;
[0154] CNN layer: used to extract local features in the input data, especially the spatial correlation of processing parameters;
[0155] LSTM layer: captures time series information in the input data to predict future knife lifting needs;
[0156] Fully connected layer: Comprehensively process the outputs of CNN and LSTM to finally get the prediction results;
[0157] Output layer: gives the knife lifting height requirement in the future time period.
[0158] The input layer input data is represented as X, which contains multiple variables, including:
[0159] Cremation gap data X1;
[0160] Electrode wear data X2;
[0161] Material property data X3;
[0162] Real-time sensor data X4;
[0163] These input data are represented as a three-dimensional tensor Where N is the number of samples, T is the time step, and D is the number of features per time step;
[0164] The convolutional neural network layer is the CNN layer, which is expressed as:
[0165] Z cnn =f(W cnn *X+b cnn )
[0166] Where: W cnn is the weight matrix of the convolution kernel; * indicates the convolution operation; b cnn is a bias term; f(·) is an activation function, such as ReLU, which can gradually extract features at different levels through multiple convolutional layers and pooling layers;
[0167] The long short-term memory network layer, or LSTM layer, consists of the following parts:
[0168] Input gate: i t =σ(W i ·[h t-1 , x t ]+b i )
[0169] Forget gate: f t =σ(W f ·[h t-1 , x t ]+b f )
[0170] Memory cell state: C t =f t ·C t-1 +i t tanh(W C ·[h t-1 , x t ]+b C )
[0171] Output gate: o t =σ(W o ·[h t-1 , x t ]+b o )
[0172] Hidden state: h t =o t tanh(C t )
[0173] Where σ(·) represents the Sigmoid activation function;
[0174] W i , W f , W C , W o is the corresponding weight matrix;
[0175] b i , b f , b C , b o is the corresponding bias term;
[0176] ·Represents bit-by-bit multiplication of elements;
[0177] The output Z of the LSTM layer lstm Will be used as input to the next fully connected layer;
[0178] The fully connected layer obtains the final prediction value:
[0179] Z fc =f(W fc ·[Z cnn , Z lstm ]+b fc )
[0180] Among them, W fc is the weight matrix of the fully connected layer;
[0181] b fc is the bias term;
[0182] f(·) is the activation function, choose ReLU or Sigmoid activation function;
[0183] The output layer gives the knife lifting height requirement in the future time period:
[0184]
[0185] Among them, W out and b out are the weights and biases of the output layer.
[0186] Furthermore, a loss function is required in the model training process, and the loss function usually uses mean square error for regression tasks:
[0187]
[0188] Among them, y iIt is the actual knife lifting height. is the predicted value of the model, and N is the number of samples.
[0189] The prediction process of the model includes the following steps:
[0190] S1: Data input: input historical processing data and real-time sensor data;
[0191] S2: Feature extraction: extract local features through CNN and capture time series information through LSTM;
[0192] S3: Comprehensive processing: The fully connected layer processes the features comprehensively to obtain the prediction results;
[0193] S4: Output prediction: The output layer gives the future prediction value of the knife lifting height.
[0194] Furthermore, the multi-sensor fusion algorithm uses the Kalman filter algorithm to integrate and optimize the data from different sensors, including two main steps:
[0195] Prediction: predict the state at the next moment based on the system's state model;
[0196] Update: Update the state estimate based on the sensor observations;
[0197] Set the system state vector x k , the observation vector is z k , the state equation and observation equation of the system are expressed as:
[0198] Equation of state:
[0199] x k =F k x k-1 +B k u k +W k
[0200] Among them: F k is the state transition matrix;
[0201] B k is the control input matrix;
[0202] u k is the control vector;
[0203] w k is the process noise, assumed to be zero-mean Gaussian white noise with covariance Q k ;
[0204] Observation equation:
[0205] z k =Hk x k +v k
[0206] Where: H k is the observation matrix;
[0207] v k is the observation noise, assumed to be zero-mean Gaussian white noise with covariance R k .
[0208] Wherein, the Kalman filtering step includes:
[0209] A1. Prediction steps:
[0210] Status prediction: x k|k-1 =F k x k-1|k-1 +B k u k
[0211] Covariance prediction: p k|k-1 =F k p k-1|k-1 F k +Q k
[0212] Where: x k|k-1 is the state prediction value at time k;
[0213] p k|k-1 is the forecast covariance matrix;
[0214] A2. Update steps:
[0215] Kalman gain calculation: K k =p k|k-1 H k (H k p k|k-1 H k +R k ) -1
[0216] Status update: x k|k =x k|k-1 +K k (z k -H k x k|k-1 )
[0217] Covariance update: p k|k =(IK k H k ) k|k-1
[0218] Among them, K kis the Kalman gain matrix, which determines the weighted ratio of prediction to observation;
[0219] x k|k is the state update value at time k;
[0220] p k|k is the updated covariance matrix.
[0221] Specific: The implementation steps of multi-sensor fusion are as follows:
[0222] In the scenario of multi-sensor fusion, assuming that there are multiple sensors providing observation data, they are integrated into the Kalman filter framework in the following way:
[0223] Multi-sensor observation:
[0224] Assume that there are M sensors, and the observation equation of each sensor is expressed as:
[0225]
[0226] Observation Fusion:
[0227] The observation data of each sensor is fused into a total observation vector:
[0228]
[0229] Among them, w i is the weighting matrix of the ith sensor, usually based on the observation noise covariance of each sensor calculate;
[0230] Observation matrix fusion:
[0231] The observation matrix H k Weighted fusion is required:
[0232]
[0233] The observation data and observation matrix after the above fusion can be directly substituted into the update step of the Kalman filter to complete the multi-sensor data fusion.
[0234] Practical application examples
[0235] In the tool lift height control of the EDM machine tool, assuming that we use a laser ranging sensor and an optical sensor to monitor the tool lift height, the system can fuse the outputs of these two sensors through Kalman filtering to obtain a more accurate tool lift height estimate.
[0236] The specific steps are as follows:
[0237] 1. Initialization: define the initial state and covariance matrix of the system;
[0238] 2. Prediction: predict the current state based on the state at the previous moment;
[0239] 3. Observation data fusion: Obtain observation data from laser ranging sensors and optical sensors and fuse them into total observation data;
[0240] 4. State update: Use the Kalman gain matrix to update the system state and obtain the optimal tool lift height estimate;
[0241] 5. Loop: Repeat the above process continuously and adjust the knife lifting height in real time.
[0242] Through Kalman filtering, the multi-sensor fusion algorithm can effectively improve the accuracy of tool lift height control and provide reliable feedback data for the artificial intelligence prediction module.
[0243] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.
Claims
1. A tool lift height control system for an electric spark machine tool, characterized in that: include: Artificial intelligence predictive control module, including: Deep learning prediction model: trained based on historical processing data, current processing parameters and real-time sensor data; the model can predict future processing status and tool lifting requirements; Data acquisition and analysis module: collects sensor data such as temperature, spark gap, current and processing data in real time, and pre-processes it for use in the prediction model; Feedback control module: According to the output of the prediction model, the tool lift height is adjusted in advance to ensure processing accuracy and stability; Energy efficiency management module, including: Energy monitoring unit: monitors the energy consumption of the knife lifting action in real time and records the power usage; Energy efficiency optimization algorithm: Combined with the results of artificial intelligence predictive control, it dynamically optimizes the tool lifting path and height to minimize energy consumption while maintaining high-precision processing; Energy efficiency feedback system: Feedback energy consumption data to the prediction control module for further optimization of the model; Human-computer interaction interface, including: Intelligent operation interface: Provides a graphical interface that enables operators to monitor and adjust system parameters in real time. The interface displays the effects of predictive adjustments and energy efficiency management. Data visualization module: provides dynamic change charts of energy consumption and tool lift height during machining to help operators optimize operation strategies; High-precision sensor fusion system, including: Laser distance sensor: used to detect spark gap and tool lift height in real time; Multi-sensor fusion algorithm: used to fuse data from multiple sensors and provide accurate feedback to the artificial intelligence prediction module.
2. The tool lift height control system for an EDM machine tool according to claim 1, characterized in that: The control method of the electric spark machine tool tool lifting height control system comprises the following steps: Step 1: Initial data collection and model training The system first collects historical data under different working conditions through multiple processing data, and then uses this data to train a deep learning model to predict possible changes and tool lifting needs in future processing; Step 2: Real-time data monitoring and prediction During the machining process, the energy monitoring unit collects the power consumption data of the EDM machine in real time, and the sensor system collects the tool lift height and spark gap data. These data are fed into the prediction model, which outputs the possible future spark gap changes and the required tool lift height. Step 3: Predictive control of tool lift height The predictive control module adjusts the tool lift height in advance based on the model's prediction results to cope with upcoming changes in working conditions. This advance adjustment can avoid a drop in machining accuracy due to sudden changes. Step 4: Implementation of energy efficiency optimization strategy Based on the output of predictive control and real-time energy monitoring results, the energy efficiency management module dynamically adjusts the tool lifting path and height to optimize energy usage; Step 5: Feedback and Adaptive Learning The system continuously learns from real-time processing data and feeds this data into the model for retraining, thereby improving prediction accuracy and energy efficiency optimization capabilities in subsequent processing.
3. The tool lift height control system for an EDM machine tool according to claim 2, characterized in that: The deep learning prediction model structure combines the feature extraction capability of CNN and the time series prediction capability of LSTM, including: Input layer: contains historical processing data such as spark gap, electrode wear, material properties and real-time sensor data such as temperature and current; CNN layer: used to extract local features in the input data, especially the spatial correlation of processing parameters; LSTM layer: captures time series information in the input data to predict future knife lifting needs; Fully connected layer: Comprehensively process the outputs of CNN and LSTM to finally get the prediction results; Output layer: gives the knife lifting height requirement in the future time period.
4. The tool lift height control system for an EDM machine tool according to claim 3, characterized in that: The input layer input data is represented as X, which contains multiple variables, including: Cremation gap data X1; Electrode wear data X2; Material property data X3; Real-time sensor data X4; These input data are represented as a three-dimensional tensor Where N is the number of samples, T is the time step, and D is the number of features per time step; The convolutional neural network layer is the CNN layer, which is expressed as: Z cnn =f(W cnn *X+b cnn ) Where: W cnn is the weight matrix of the convolution kernel; * indicates the convolution operation; b cnn is a bias term; f(·) is an activation function, such as ReLU, which can gradually extract features at different levels through multiple convolutional layers and pooling layers; The long short-term memory network layer, or LSTM layer, consists of the following parts: Input gate: i t =σ(W i ·[h t-1 ,x t ]+b i ) Forget gate: f t =σ(W f ·[h t-1 ,x t ]+b f ) Memory cell state: C t =f t ·C t-1 +i t tanh(W C ·[h t-1 ,x t ]+b C ) Output gate: o t =σ(W o ·[h t-1 ,x t ]+b o ) Hidden state: h t =o t tanh(C t ) Where σ(·) represents the Sigmoid activation function; W i , W f , W C , W o is the corresponding weight matrix; b i , b f , b C , b o is the corresponding bias term; ·Represents bit-by-bit multiplication of elements; The output Z of the LSTM layer lstm Will be used as input to the next fully connected layer; The fully connected layer obtains the final prediction value: WITH fc =f(W fc ·[WITH cnn ,WITH lstm ]+b fc ) Among them, W fc is the weight matrix of the fully connected layer; b fc is the bias term; f(·) is the activation function, choose ReLU or Sigmoid activation function; The output layer gives the knife lifting height requirement in the future time period: Among them, W out and b out are the weights and biases of the output layer.
5. The tool lift height control system for an EDM machine tool according to claim 4, characterized in that: The loss function is required in the model training process. The loss function usually uses mean square error for regression tasks: Among them, y i It is the actual knife lifting height. is the predicted value of the model, and N is the number of samples.
6. The tool lift height control system for an EDM machine tool according to claim 5, characterized in that: The prediction process of the model includes the following steps: S1: Data input: input historical processing data and real-time sensor data; S2: Feature extraction: extract local features through CNN and capture time series information through LSTM; S3: Comprehensive processing: The fully connected layer processes the features comprehensively to obtain the prediction results; S4: Output prediction: The output layer gives the future prediction value of the knife lifting height.
7. The tool lift height control system for an EDM machine tool according to claim 1, characterized in that: The multi-sensor fusion algorithm uses the Kalman filter algorithm to integrate and optimize the data from different sensors, including two main steps: Prediction: predict the state at the next moment based on the system's state model; Update: Update the state estimate based on the sensor observations; Set the state vector xk of the system , The observation vector is z k , the state equation and observation equation of the system are expressed as: Equation of state: x k =F k x k-1 +B k u k +w k Among them: F k is the state transition matrix; B k is the control input matrix; u k is the control vector; w k is the process noise, assumed to be zero-mean Gaussian white noise with covariance Q k ; Observation equation: z k =H k x k +v k Where: H k is the observation matrix; v k is the observation noise, assumed to be zero-mean Gaussian white noise with covariance R k .
8. The tool lift height control system for an EDM machine tool according to claim 7, characterized in that: The Kalman filtering step comprises: A1. Prediction steps: Status prediction: x k|k-1 =F k x k-1|k-1 +B k u k Covariance prediction: p k|k-1 =F k p k-1|k-1 F k +Q k Where: x k|k-1 is the state prediction value at time k; p k|k-1 is the forecast covariance matrix; A2. Update steps: Kalman gain calculation: K k =p k|k-1 H k (H k p k|k-1 H k +R k ) -1 Status update: x k|k =x k|k-1 +K k (z k -H k x k|k-1 ) Covariance update: p k|k =(IK k H k ) k|k-1 Among them, K k is the Kalman gain matrix, which determines the weighted ratio of prediction to observation; x k|k is the state update value at time k; p k|k is the updated covariance matrix.
9. The tool lift height control system for an EDM machine tool according to claim 8, characterized in that: The implementation steps of multi-sensor fusion are as follows: In the scenario of multi-sensor fusion, assuming that there are multiple sensors providing observation data, they are integrated into the Kalman filter framework in the following way: Multi-sensor observation: Assume that there are M sensors, and the observation equation of each sensor is expressed as: Observation Fusion: The observation data of each sensor is fused into a total observation vector: Among them, w i is the weighting matrix of the ith sensor, usually based on the observation noise covariance of each sensor calculate; Observation matrix fusion: The observation matrix H k Weighted fusion is required: The observation data and observation matrix after the above fusion can be directly substituted into the update step of the Kalman filter to complete the multi-sensor data fusion.
Citation Information
Cited By
Workpiece detection method, spark machine, computer readable storage medium and program product
CN119772282A
