Digital mechanical equipment safety distance analysis system based on artificial intelligence

Abstract

The invention discloses a digital mechanical equipment safety distance analysis system based on artificial intelligence, and the system obtains equipment operation parameters and surrounding environment three-dimensional information through multi-dimensional data collection and binocular vision image collection, carries out the safety distance analysis through an optimized support vector machine model after preprocessing and feature extraction, and achieves the safety distance analysis of the equipment. Equipment parameters such as load weight and temperature are introduced into model training to construct an influence model, and adaptive noise reduction based on operation frequency and a compensation model based on vibration parameters are adopted in binocular vision processing. The system generates an early warning signal and a control instruction according to an analysis result, realizes intelligent adjustment of the equipment operation state, effectively overcomes the defects of weak data processing capability and lack of intelligent decision making in the traditional technology through multi-unit cooperation and algorithm optimization, remarkably improves the accuracy of safety distance analysis and system adaptability, and improves the safety distance analysis efficiency. And safe operation of digital mechanical equipment is ensured.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of mechanical equipment safety distance analysis, and particularly relates to a digital mechanical equipment safety distance analysis system based on artificial intelligence. BACKGROUND

[0002] At present, industrial automation and intelligent manufacturing are developing rapidly, and digital mechanical equipment is widely used in various production scenarios. The safety problem in the operation process of the equipment is crucial. The control of the safety distance between the equipment and the surrounding objects and personnel during operation is a key factor to ensure production safety and avoid accidents. With the increasing complexity of industrial environments, traditional manual monitoring and simple sensor warning methods have been difficult to meet the high-precision and real-time safety distance analysis requirements, and there is an urgent need for advanced technology to build an intelligent safety distance analysis system.

[0003] The existing technology based on mechanical equipment safety distance analysis has obvious shortcomings. Firstly, the data processing and analysis capability is limited. Traditional systems rely on single sensors to collect data, which cannot fully obtain the equipment operating state and surrounding environment information, and the data processing algorithm is simple, which makes it difficult to mine the correlation between complex data, resulting in low safety distance analysis accuracy and inability to accurately respond to dynamic changes in working conditions. Secondly, it lacks intelligent decision-making and adaptive ability. Most existing technologies only make safety judgments based on preset thresholds, cannot flexibly adjust safety strategies according to changes in equipment parameters and environmental differences, and are prone to misjudgment or omission in complex working conditions, making it difficult to ensure safe operation of the equipment. SUMMARY

[0004] In order to overcome the shortcomings and deficiencies of the existing technology, the present application provides a digital mechanical equipment safety distance analysis system based on artificial intelligence.

[0005] The technical scheme adopted by the present application is a digital mechanical equipment safety distance analysis system based on artificial intelligence, which comprises:

[0006] A multi-dimensional data acquisition and fusion unit is arranged at a key part of the digital mechanical equipment to collect spatial position, speed, acceleration and component angle parameters during equipment operation, integrate multi-source data to form an original set using data fusion technology, and transmit the original set to a data preprocessing and feature extraction unit;

[0007] A data preprocessing and feature extraction unit receives the original data set, eliminates dimensional differences through standardization processing, extracts a feature vector related to safety distance analysis using a preset algorithm, and then sends the feature vector to a binocular vision image acquisition and processing unit;

[0008] Binocular vision image acquisition and processing unit: two sets of specific focal length and field angle vision sensors are configured to collect peripheral environment image data, which is processed by grayscale and noise reduction, and the three-dimensional space information is obtained by calculating the disparity through the stereo matching algorithm, and is transmitted to the optimized support vector machine training and decision unit;

[0009] Optimized support vector machine training and decision unit: receiving feature vectors and three-dimensional space information, based on the optimized support vector machine algorithm, training the model using historical data, inputting real-time data into the trained model to output safety distance decision information, and sending to the safety distance early warning and control instruction generation unit;

[0010] Safety distance early warning and control instruction generation unit: receiving decision information, judging the equipment running state according to the preset threshold, generating a warning signal when it is unsafe, and generating a control instruction combined with the equipment control logic, and transmitting to the early warning and control execution unit;

[0011] Early warning and control execution unit: receiving warning signals and control instructions, starting the sound and light alarm warning device, adjusting the equipment running speed, direction or starting and stopping according to the instructions.

[0012] Further, the optimized support vector machine training and decision unit adopts the following optimized support vector machine model for safety distance analysis:

[0013]

[0014] Wherein, D is the safety distance value calculated between the mechanical equipment and the surrounding objects; n is the number of training samples; α i is the Lagrange multiplier corresponding to the i-th training sample, whose value range is determined by the optimization algorithm; y i is the class label of the i-th training sample, representing whether it is within the safety distance range; K(x i , x) is the kernel function, x i is the feature vector of the i-th training sample, x is the current input feature vector to be analyzed, and the kernel function is used to map the data in the low-dimensional space to the high-dimensional space for better classification; b is the offset of the classification hyperplane, which is determined through the training process.

[0015] Further, the binocular vision image acquisition and processing unit adopts the following disparity calculation model in the stereo matching process:

[0016]

[0017] Wherein, d is the calculated disparity; f is the focal length of the visual sensor, which is an inherent parameter of the binocular visual sensor; B is the baseline distance between the two visual sensors, which is determined when the system is installed; Z is the actual distance of the object to the visual sensor; At the same time, the calculated disparity is corrected in combination with the spatial position parameter P(x, y, z) of the digitized mechanical equipment, and the corrected disparity d 修正 The calculation formula is:

[0018] d 修正 = d + Ad(P(x, y, z))

[0019] Wherein, Ad(P(x, y, z)) is the disparity correction amount based on the spatial position parameter P(x, y, z) of the mechanical equipment.

[0020] Further, the optimized support vector machine training and decision unit, in the model training process, the load weight W of the mechanical equipment is introduced as the influencing factor, and the following improved optimization objective function is constructed:

[0021]

[0022] s.t.y i (ω T φ(x i )+b)≥1-ξ i , ξ i ≥0, i = 1, …, n

[0023] Wherein, ω is the normal vector of the hyperplane; b is the offset of the classification hyperplane; ξ i is the relaxation variable of the i-th sample; C is a penalty parameter, which is used to balance the complexity of the model and the classification error; λ is the influence coefficient of the load weight W; φ(x i ) is the result of mapping the input sample x i to high-dimensional space through the kernel function.

[0024] Further, the binocular visual image acquisition and processing unit, when carrying out the noise reduction processing on the collected image, adopts an adaptive noise reduction model based on the running frequency f r of the mechanical equipment:

[0025]

[0026] Wherein, I 降噪 (x, y) is the pixel value of the denoised image at coordinates (x, y); I(m, n) is the pixel value of the original image at coordinates (m, n); N(x, y) is the neighborhood centered at coordinates (x, y); f(x, y) is the frequency feature of the pixel at coordinates (x, y); β is a parameter for controlling the weight decay speed.

[0027] Further, the safety distance warning and control instruction generation unit is based on the safety distance decision information D output by the optimized support vector machine and the braking performance parameter B of the mechanical equipment p , a control instruction generation model is constructed:

[0028]

[0029] Wherein, C 指令 is the generated control instruction, D 紧急 is the emergency braking safety distance threshold, C 警告 is the safety distance warning threshold, and the two thresholds are determined according to the braking performance parameter B p of the mechanical equipment.

[0030] Further, the optimized support vector machine training and decision unit, in the model training, combined with the influence of temperature parameter T of mechanical equipment on safety distance analysis, constructs temperature correction model:

[0031] D 修正 =D×(1+γ(T-T0))

[0032] Wherein, D 修正 is the corrected safety distance value; D is the safety distance value calculated by the optimized support vector machine model initially; γ is the temperature influence coefficient; T is the current temperature parameter of the mechanical equipment; T0 is the standard temperature reference value.

[0033] Further, the binocular vision image acquisition and processing unit, in the stereoscopic matching to obtain three-dimensional space information, combined with the vibration parameter V of the mechanical equipment, constructs the vibration compensation model:

[0034] Z 补偿 =Z+ΔZ(V)

[0035] Wherein, Z 补偿 is the compensated actual distance from the object to the vision sensor, is the original distance calculated by the parallax, and ΔZ(V) is the distance compensation amount based on the vibration parameter V of the mechanical equipment, which is determined by the pre-established vibration-distance compensation relationship model.

[0036] Further, the optimized support vector machine training and decision unit, in the model training process, combined with the hydraulic pressure parameter P h of the mechanical equipment, constructs a hydraulic pressure influence model: Wherein, is the Lagrange multiplier corresponding to the i-th training sample after the correction of the influence of the hydraulic pressure; α i is the original Lagrange multiplier; μ is the hydraulic pressure influence coefficient; P hP is a current hydraulic pressure parameter of the mechanical equipment h0 P is a standard hydraulic pressure reference value.

[0037] An artificial intelligence-based digital mechanical equipment safety distance analysis system, the system operation includes the following steps:

[0038] Step one: using the multi-dimensional data acquisition and fusion unit set at the key parts of the digital mechanical equipment, collecting the spatial position parameters, running speed parameters, acceleration parameters, and equipment component angle parameters multi-source data during equipment operation, and fusing the collected data to form a multi-dimensional original data set;

[0039] Step two: the original data set obtained by the multi-dimensional data acquisition and fusion unit is transmitted to the data preprocessing and feature extraction unit, the original data is standardized, and the preset feature extraction algorithm is used to extract the feature vector related to the mechanical equipment safety distance analysis from the standardized data;

[0040] Step three: through the binocular vision image acquisition and processing unit equipped with two groups of specific focal length and field of view angle visual sensors, the surrounding environment of the mechanical equipment is image collected, the binocular vision image data is obtained, the collected image data is grayed and denoised, the parallax of each point in the image is calculated through the stereo matching algorithm, and then the three-dimensional space information of the surrounding environment is obtained;

[0041] Step four: the feature vector extracted by the data preprocessing and feature extraction unit and the three-dimensional space information obtained by the binocular vision image acquisition and processing unit are transmitted to the optimized support vector machine training and decision unit, based on the optimized support vector machine algorithm, the support vector machine model is trained using historical data, and a safety distance analysis model is constructed. The real-time extracted feature vector and three-dimensional space information are input into the trained safety distance analysis model, and the safety distance decision information of the mechanical equipment and the surrounding objects is output;

[0042] Step five: the safety distance decision information output by the optimized support vector machine training and decision unit is transmitted to the safety distance warning and control instruction generation unit, according to the preset safety distance threshold, it is judged whether the current running state of the mechanical equipment is in the safety range, when the unsafe state is detected, the corresponding warning signal is generated, and according to the safety distance decision information and the control logic of the equipment, the control instruction is generated;

[0043] Step six: the warning signal and control instruction generated by the safety distance warning and control instruction generation unit are transmitted to the warning and control execution unit, according to the warning signal, the corresponding warning device is started, and according to the control instruction, the running state of the digital mechanical equipment is adjusted.

[0044] Beneficial effects: The present invention proposes a digital mechanical equipment safety distance analysis system based on artificial intelligence. The system integrates the spatial position, speed, acceleration and other parameters of the equipment operation through a multi-dimensional data acquisition and fusion unit, and combines the three-dimensional information of the surrounding environment obtained by the binocular vision image acquisition and processing unit to solve the problem of single data acquisition in traditional technologies and provide a comprehensive data basis for accurate analysis. At the data processing level, the data preprocessing and feature extraction unit standardizes the multi-source data and extracts key features. The optimized support vector machine training and decision-making unit constructs a safety distance analysis model based on the optimization algorithm, deeply mines data associations, and greatly improves the accuracy of safety distance analysis compared to traditional simple algorithms. In view of the shortcomings of traditional technologies in lacking intelligent decision-making and adaptive capabilities, the system introduces equipment parameters such as load weight, temperature, and hydraulic pressure to construct an influence model during the model training and decision-making process, and dynamically adjusts the safety distance analysis strategy according to the real-time status of the equipment; the binocular vision image acquisition and processing unit combines the equipment operating frequency and vibration parameters to establish an adaptive noise reduction and vibration compensation model to ensure the acquisition of accurate environmental information, so that the system can flexibly respond to complex working conditions and avoid misjudgment and missed judgments. The safety distance warning and control instruction generation unit generates precise control instructions based on the optimized support vector machine output and the equipment's braking performance parameters. The warning and control execution unit responds promptly, realizing intelligent safety management and control, ensuring the safe and stable operation of digital mechanical equipment, and significantly improving the safety and reliability of industrial production. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 It is a diagram of the system unit composition of the present invention;

[0046] Figure 2 It is a flow chart of the system operation steps of the present invention. DETAILED DESCRIPTION

[0047] It should be noted that, unless there is a conflict, the embodiments in this application and the features described in the embodiments can be combined with each other. The application is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0048] like Figure 1 As shown, a digital mechanical equipment safety distance analysis system based on artificial intelligence includes:

[0049] Multi-dimensional data acquisition and fusion unit: This unit is installed in the key parts of digital mechanical equipment and is used to collect various parameters of the mechanical equipment during operation, including but not limited to the spatial position parameters, operating speed parameters, acceleration parameters and angular parameters of each component of the equipment. The collected multi-source data is integrated through data fusion technology to form a multi-dimensional raw data set, which is then transmitted to the data preprocessing and feature extraction unit;

[0050] Specifically, the unit is deployed at the key parts of the digitized mechanical equipment, and its core function is to collect all kinds of parameters in the running process of the equipment in all directions and in real time. These parameters include spatial position parameters of the equipment, which accurately record the coordinate information of the equipment in three-dimensional space; running speed parameters, which monitor the moving speed of the equipment in real time; acceleration parameters, which reflect the speed change of the equipment; and angle parameters of each part of the equipment, which are used to master the attitude of each part of the equipment. Through multi-dimensional data collection, the unit can comprehensively obtain the running state information of the equipment. In the implementation, the unit is equipped with a variety of high-precision sensors, which accurately measure different parameters to ensure that the collected data is accurate and reliable. Then, the unit uses advanced data fusion technology to integrate and process the collected multi-source heterogeneous data, eliminate the redundancy and contradiction between the data, and form a complete and unified original data set. This set provides comprehensive and basic data support for subsequent safety distance analysis and is the cornerstone of the accurate operation of the whole system.

[0051] The technical parameters of the unit include the precision of the sensor, the sampling frequency, and the complexity of the data fusion algorithm. High-precision sensors can ensure that the collected data has minimal error, and the sampling frequency determines the real-time nature of the data. The higher the frequency, the more timely the changes in the running state of the equipment can be reflected. The complexity of the data fusion algorithm affects the quality and reliability of the integrated data. The significance of the unit is to provide the system with comprehensive and accurate original data, so that subsequent analysis and decision-making can be based on a solid data foundation. Without the effective work of the unit, the subsequent safety distance analysis will be a sourceless water and cannot accurately reflect the actual running condition of the equipment and the surrounding environment.

[0052] Data preprocessing and feature extraction unit: receives the original data set from the multi-dimensional data collection and fusion unit, standardizes the original data, eliminates the dimensional differences of the data, uses a pre-set feature extraction algorithm to extract the feature vector related to the safety distance analysis of the mechanical equipment from the standardized data, and sends the feature vector to the binocular vision image collection and processing unit;

[0053] Specifically, this unit receives the raw data set from the multi-dimensional data acquisition and fusion unit, and its primary task is to standardize the raw data. Since the data collected by different sensors may have different dimensions and magnitudes, standardization can eliminate these differences and make the data in the same dimension system, which is convenient for subsequent analysis and processing. In the implementation process, the unit will select the appropriate standardization method according to the characteristics of the data, such as Z-score standardization, Min-Max standardization, etc., to ensure the consistency and comparability of the data. Subsequently, the unit uses the pre-set feature extraction algorithm to mine the feature vectors closely related to the safety distance analysis of mechanical equipment from the standardized data. These feature vectors can highlight the essential features and internal relations of the data, providing key evidence for subsequent model training and decision-making.

[0054] The technical parameters of this unit include the selection of standardization methods, the type of feature extraction algorithm, and the dimension of feature vectors, etc. Different standardization methods are suitable for different types of data, and selecting the appropriate method can improve the quality of data and analysis effect. The type of feature extraction algorithm directly affects the effectiveness and representativeness of the extracted feature vectors, and common algorithms include principal component analysis, wavelet transform, etc. The dimension of the feature vector needs to be reasonably selected according to the actual situation, and too high or too low dimension may affect the accuracy of the analysis result. The significance of this unit is to clean and refine the raw data, remove useless information, extract key features, thereby reducing the computational complexity of subsequent processing, improving the analysis efficiency and accuracy of the system. It is an important link for data to transform from the original state to usable information.

[0055] Binocular vision image acquisition and processing unit: equipped with two groups of vision sensors with specific focal length and field of view, image acquisition of the surrounding environment of mechanical equipment, acquisition of binocular vision image data, processing of collected image data such as grayscale and noise reduction, calculation of the disparity of each point in the image through stereo matching algorithm, and then acquisition of the three-dimensional space information of the surrounding environment, and transmission of the three-dimensional space information to the optimized support vector machine training and decision-making unit;

[0056] Specifically, the unit is equipped with two sets of visual sensors with specific focal lengths and field of view angles, specifically for image acquisition of the mechanical equipment's surrounding environment. In implementation, the two sets of visual sensors are installed at precise positions and angles to ensure comprehensive and clear image data of the surrounding environment. The collected image data is first subjected to grayscale processing, converting color images into grayscale images to reduce data volume while retaining the main features of the images. Then, noise reduction processing is performed to remove noise in the images caused by factors such as light and equipment jitter, improving the quality of the images. Then, the stereo matching algorithm is used to calculate the disparity of each point in the image, and thus obtain the three-dimensional spatial information of the surrounding environment. In this process, the algorithm accurately matches the corresponding points in the left and right images, calculates the disparity between them, and then calculates the three-dimensional spatial position of the object based on the relationship between disparity and distance.

[0057] The technical parameters of this unit include the focal length, field of view angle, resolution of the visual sensor, and the accuracy of the stereo matching algorithm. The focal length and field of view angle determine the range and distance that the sensor can capture, while the resolution affects the clarity and detail of the image. The accuracy of the stereo matching algorithm directly affects the accuracy of the three-dimensional spatial information obtained, and a high-precision algorithm can more accurately calculate the disparity and thus obtain more accurate object position information. The significance of this unit lies in providing the system with intuitive and comprehensive surrounding environment information, enabling the system to real-time perceive the distribution and distance of objects around the equipment. These three-dimensional spatial information are crucial for accurately analyzing the safety distance of the mechanical equipment and are the key basis for the system to achieve precise early warning and control.

[0058] Optimized support vector machine training and decision-making unit: receives the feature vectors from the data preprocessing and feature extraction unit and the three-dimensional spatial information from the binocular vision image acquisition and processing unit, trains the support vector machine model based on the optimized support vector machine algorithm using historical data, constructs a safety distance analysis model, inputs the real-time acquired feature vectors and three-dimensional spatial information into the trained safety distance analysis model, outputs the safety distance decision information of the mechanical equipment and surrounding objects, and sends the decision information to the safety distance early warning and control instruction generation unit;

[0059] Specifically, this unit receives the feature vector from the data preprocessing and feature extraction unit and the three-dimensional spatial information from the binocular vision image acquisition and processing unit. Its core task is to perform safety distance analysis based on an optimized support vector machine algorithm. In the implementation process, the unit first trains the support vector machine model using historical data. These historical data contain the feature vectors of the equipment in various operating states and the corresponding safety distance conditions. Through learning from these data, the model can grasp the internal relationship between the feature vector and the safety distance. To improve the accuracy and adaptability of the model, the unit uses an optimized support vector machine algorithm to continuously adjust the parameters of the model during training, so that it can better fit the data. After training, the real-time acquired feature vector and three-dimensional spatial information are input into the trained model, and the model outputs the safety distance decision information of the mechanical equipment and the surrounding objects.

[0060] The technical parameters of this unit include the kernel function type of the support vector machine model, the penalty parameter, and the selection of the optimization algorithm. Different kernel functions are suitable for different types of data, and selecting the appropriate kernel function can improve the classification performance of the model. The penalty parameter controls the error tolerance of the model during training and needs to be reasonably adjusted according to the actual situation. The selection of the optimization algorithm directly affects the efficiency and quality of model training, and common optimization algorithms include the SMO algorithm. The significance of this unit lies in analyzing and learning a large amount of data through advanced machine learning algorithms, thereby achieving accurate prediction and judgment of the safety distance of mechanical equipment. It is the intelligent core of the system and can provide scientific and reliable decision-making basis for subsequent early warning and control.

[0061] Safety distance early warning and control instruction generation unit: receives the safety distance decision information from the optimized support vector machine training and decision unit, judges whether the current operating state of the mechanical equipment is within the safety range according to the preset safety distance threshold, generates the corresponding early warning signal when an unsafe state is detected, and generates the control instruction according to the safety distance decision information and the control logic of the equipment, and transmits the early warning signal and the control instruction to the early warning and control execution unit;

[0062] Specifically, the unit receives safety distance decision information from the optimized support vector machine training and decision unit, and its main function is to judge the current operating state of the mechanical equipment according to the preset safety distance threshold. In implementation, the unit compares the received safety distance decision information with the pre-set safety distance threshold of different levels. If the safety distance is less than or equal to the emergency braking threshold, it means that the equipment is in a very dangerous state, and the unit will immediately generate a control instruction for emergency braking; if the safety distance is between the emergency braking threshold and the warning threshold, the unit will generate a deceleration instruction to remind the equipment to reduce the operating speed; if the safety distance is greater than the warning threshold, it is considered that the equipment is in a safe state, and no control instruction is generated. At the same time, when an unsafe state is detected, the unit will also generate a corresponding warning signal to notify relevant personnel to take measures in time.

[0063] The technical parameters of this unit include the setting of safety distance threshold, the type of warning signal and the generation logic of control instruction, etc. The setting of safety distance threshold needs to be reasonably adjusted according to the type of equipment, operating environment and safety requirements, etc., to ensure that the safe operation of the equipment is guaranteed, and the work efficiency of the equipment is not excessively limited. The type of warning signal can include sound and light alarm, SMS notification, etc., and different types are suitable for different scenarios. The generation logic of control instruction needs to be designed in combination with the control logic and safety strategy of the equipment to ensure the rationality and effectiveness of the instruction. The significance of this unit is to convert abstract safety distance decision information into specific and executable warning signals and control instructions, providing direct protection for the safe operation of the equipment. It is the key transitional link from analysis to execution of the system.

[0064] Warning and control execution unit: receives warning signals and control instructions from the safety distance warning and control instruction generation unit, starts the corresponding warning device such as sound and light alarm equipment according to the warning signal, and adjusts the operating state of the digital mechanical equipment such as controlling the speed, direction or start-stop of the equipment according to the control instruction. The above units realize data interaction and transmission through data transmission lines or communication protocols.

[0065] Specifically, the unit receives warning signals and control instructions from the safety distance warning and control instruction generation unit, and its core task is to convert these signals and instructions into actual actions. In the implementation process, when receiving a warning signal, the unit will immediately start the corresponding warning device. For example, if it is an audible and visual alarm device, the unit will control the device to emit a loud alarm sound and a prominent flashing light to attract the attention of relevant personnel. For control instructions, the unit will accurately adjust the running state of the digital mechanical equipment according to the content of the instructions. If it is an emergency braking instruction, the unit will quickly trigger the braking system of the equipment to make it stop running immediately; if it is a speed reduction instruction, the unit will adjust the power output of the equipment to reduce the running speed of the equipment. The entire response process is quick and accurate, ensuring that various safety situations can be responded to in a timely manner.

[0066] The technical parameters of the unit include the response time of the warning device, the execution accuracy of the control instruction, and the stability of the system, etc. The shorter the response time of the warning device, the more timely it can remind personnel to take measures to reduce the possibility of accidents. The execution accuracy of the control instruction directly affects the effect of adjusting the running state of the equipment, and high-precision execution can ensure that the equipment accurately changes the running state according to the instruction. The stability of the system ensures that the unit does not fail during long-term operation and can work continuously and reliably. The significance of the unit is to convert the analysis and decision results of the system into actual actions, which is the final execution link of the entire safety distance analysis system. Without the effective work of the unit, all the efforts of the previous units will be in vain, and the safety protection function of the system cannot be realized.

[0067] Preferably, the optimized support vector machine training and decision unit adopts the following optimized support vector machine model for safety distance analysis:

[0068]

[0069] where D is the calculated safety distance value of the mechanical equipment and the surrounding object; n is the number of training samples; a i is the Lagrange multiplier corresponding to the i-th training sample, whose value range is determined by the optimization algorithm; y i is the class label of the i-th training sample, representing whether it is within the safety distance range; K(x i , x) is a kernel function, x i is the feature vector of the i-th training sample, x is the current input feature vector to be analyzed, and the kernel function is used to map the data in the low-dimensional space to the high-dimensional space for better classification; b is the offset of the classification hyperplane, which is determined through the training process; the feature vector extracted by the data preprocessing and feature extraction unit includes the running speed v, acceleration a, and equipment component angle θ of the mechanical equipment, which constitute the feature vectors x and xi part of the composition, the training and decision of the optimized support vector machine model.

[0070] Specifically, the safety distance analysis is realized by the optimized support vector machine model, and the model constructs a calculation framework with the number of training samples, the Lagrange multiplier, the category label, the kernel function and the offset as parameters. The value range of the Lagrange multiplier is determined by the optimization algorithm, and the kernel function maps the low-dimensional data to the high-dimensional space to improve the classification effect. The parameters such as running speed, acceleration and device component angle extracted by the data preprocessing and feature extraction unit constitute the feature vector, which is used for model training and decision, and the precise calculation of the safety distance of the mechanical equipment and the surrounding objects is realized through the model.

[0071] Preferably, the binocular vision image acquisition and processing unit adopts the following disparity calculation model in the stereo matching process:

[0072]

[0073] Wherein, d is the calculated disparity; f is the focal length of the vision sensor, which is the inherent parameter of the binocular vision sensor; B is the baseline distance between the two vision sensors, which is determined during system installation; Z is the actual distance of the object to the vision sensor; At the same time, the calculated disparity is corrected combined with the spatial position parameter P(x, y, z) of the digital mechanical equipment, and the corrected disparity d 修正 The calculation formula is:

[0074] d 修正 = d + Ad(P(x, y, z))

[0075] Wherein, Ad(P(x, y, z)) is the disparity correction amount based on the spatial position parameter P(x, y, z) of the mechanical equipment, which is determined by the pre-established correction model, which combines the influence of the position of the mechanical equipment itself on the calculation of the disparity of the surrounding objects, and is used to more accurately obtain the three-dimensional space information of the surrounding environment and transmit it to the optimized support vector machine training and decision unit.

[0076] Specifically, the binocular vision image acquisition and processing unit introduces the disparity calculation model in the stereo matching process, calculates the disparity with the focal length of the vision sensor, the baseline distance and the actual distance of the object, and corrects the disparity combined with the spatial position parameter of the mechanical equipment. The correction amount is determined by the pre-established model considering the influence of the position of the device itself on the disparity, so as to more accurately obtain the three-dimensional space information of the surrounding environment and provide reliable data basis for the subsequent optimized support vector machine training and decision unit.

[0077] Preferably, the optimized support vector machine training and decision unit, in the model training process, the load weight W of the mechanical equipment is introduced as an influencing factor, and the following improved optimization objective function is constructed:

[0078]

[0079] s.t.y i (ω T φ(x i )+b)≥1-ξ i ,ξ i ≥0,i=1,…,n

[0080] Wherein, ω is the normal vector of the hyperplane; b is the offset of the classification hyperplane; ξ i is the relaxation variable of the ith sample; C is the penalty parameter, which is used to balance the complexity of the model and the classification error; λ is the influence coefficient of the load weight W; φ(x i ) is the result of mapping the input sample x i to a high-dimensional space through a kernel function; in the training process, by adjusting the above parameters, combined with the feature vector containing the load weight W and other parameters extracted by the data preprocessing and feature extraction unit, the support vector machine model is optimized, the accuracy of the safety distance analysis is improved, and the trained model is used for safety distance decision of real-time input data.

[0081] Specifically, in the model training of the optimized support vector machine training and decision unit, the load weight is introduced as an influencing factor to construct an improved optimization objective function. The function takes the hyperplane normal vector, offset, relaxation variable, penalty parameter and load weight influence coefficient as parameters, optimizes the model by adjusting these parameters and combining the feature vector containing the load weight and other parameters, effectively improves the accuracy of safety distance analysis, and makes the model better adapt to safety distance decision under different load conditions.

[0082] Preferably, the binocular vision image acquisition and processing unit, when carrying out noise reduction processing on the collected image, adopts an adaptive noise reduction model based on the running frequency f r of the mechanical equipment:

[0083]

[0084] Wherein, I 降噪 (x, y) is the pixel value of the noise-reduced image at coordinates (x, y); I(m, n) is the pixel value of the original image at coordinates (m, n); N(x, y) is the neighborhood centered at coordinates (x, y); f(x, y) is the frequency feature of the pixel at coordinates (x, y); β is a parameter for controlling the weight decay speed; through the adaptive noise reduction model, combined with the running frequency f rPerform noise reduction processing on the image to reduce the impact of interference noise generated by equipment operation on image quality, improve the accuracy of subsequent stereo matching and three-dimensional spatial information acquisition, and use the processed image information to calculate the three-dimensional spatial information of the surrounding environment.

[0085] Specifically, an adaptive noise reduction model based on the operating frequency of mechanical equipment is used in the image noise reduction process of the binocular vision image acquisition and processing unit. This model constructs a calculation system based on the denoised pixel values, original pixel values, neighborhood range, pixel frequency characteristics, and parameters that control the weight decay rate. By combining the operating frequency of the equipment to perform image noise reduction, it reduces the impact of interference noise generated by the equipment operation on image quality, thereby improving the accuracy of subsequent stereo matching and three-dimensional spatial information acquisition.

[0086] Preferably, the safety distance warning and control instruction generating unit is based on the safety distance decision information D output by the optimized support vector machine and the braking performance parameter B of the mechanical equipment. p , build a control instruction generation model:

[0087]

[0088] Among them, C 指令 is the generated control instruction, D 紧急 is the emergency braking safety distance threshold, C 警告 The safety distance warning thresholds are based on the braking performance parameters B of the mechanical equipment. p , determined by a pre-established relationship model, which combines the influence of factors such as equipment braking distance and braking time on the safety distance threshold, and is used to accurately generate corresponding control instructions based on the safety distance decision information, and transmit the control instructions to the early warning and control execution unit.

[0089] Specifically, the safety distance warning and control command generation unit constructs a control command generation model based on the safety distance decision information output by an optimized support vector machine and the mechanical equipment's braking performance parameters. This model categorizes safety distances into different levels based on preset emergency braking safety distance thresholds and safety distance warning thresholds, generating corresponding emergency braking, deceleration, or no-operation commands. The thresholds are determined using a relationship model that considers factors such as the equipment's braking distance and braking time, ensuring accurate control command generation.

[0090] Preferably, the optimized support vector machine training and decision-making unit, during model training, combines the influence of the temperature parameter T of the mechanical equipment on the safety distance analysis to construct a temperature correction model:

[0091] D 修正 =D×(1+γ(T-T0))

[0092] Among them, D修正 Dcorrected = D + γ (T - T0), where Dcorrected is the corrected safety distance value; D is the safety distance value calculated by the optimized support vector machine model initially; γ is the temperature influence coefficient; T is the current temperature parameter of the mechanical equipment; T0 is the standard temperature reference value. Through the temperature correction model, the safety distance output by the support vector machine model is corrected by combining the temperature parameter T extracted by the data preprocessing and feature extraction unit, the accuracy of the safety distance analysis result under different temperature environments is improved, and the corrected safety distance decision information is sent to the safety distance early warning and control instruction generation unit.

[0093] Specifically, during the model training of the optimized support vector machine training and decision unit, the influence of the temperature parameter of the mechanical equipment on the safety distance analysis is considered, and a temperature correction model is constructed. The model corrects the safety distance output by the support vector machine model by taking the initial safety distance value, the temperature influence coefficient, the current temperature parameter, and the standard temperature reference value as parameters, improves the analysis accuracy of the model under different temperature environments, and makes the safety distance decision information more consistent with the actual working condition.

[0094] Preferably, the binocular vision image acquisition and processing unit, when acquiring three-dimensional space information through stereo matching, combines the vibration parameter V of the mechanical equipment to construct a vibration compensation model:

[0095] Z 补偿 = Z + ΔZ(V)

[0096] wherein Z 补偿 is the actual distance from the object to the vision sensor after compensation, Z is the original distance calculated through parallax, and ΔZ(V) is the distance compensation amount based on the vibration parameter V of the mechanical equipment, which is determined through a pre-established vibration-distance compensation relationship model. The model combines the influence of factors such as equipment vibration amplitude and frequency on object distance measurement, is used for compensating and correcting the distance parameter in the acquired three-dimensional space information, improves the accuracy of the three-dimensional space information, and transmits the corrected three-dimensional space information to the optimized support vector machine training and decision unit.

[0097] Specifically, during the process of acquiring three-dimensional space information through stereo matching of the binocular vision image acquisition and processing unit, a vibration compensation model is constructed considering the vibration parameter of the mechanical equipment. The model takes the original distance and the distance compensation amount based on the vibration parameter as parameters, determines the compensation amount through a pre-established relationship model considering the influence of factors such as equipment vibration amplitude and frequency on object distance measurement, compensates and corrects the distance parameter in the acquired three-dimensional space information, and improves the accuracy of the environmental information.

[0098] Preferably, the optimized support vector machine training and decision unit, during the model training process, combines the hydraulic pressure parameter P h of the mechanical equipment to construct a hydraulic pressure influence model: wherein, is the i-th training sample corresponding to the Lagrange multiplier after the hydraulic pressure influence correction; a i is the original Lagrange multiplier; μ is the hydraulic pressure influence coefficient; P h is the current hydraulic pressure parameter of the mechanical equipment; P h0 is the standard hydraulic pressure reference value. Through the hydraulic pressure influence model, in the support vector machine model training process, according to the hydraulic pressure parameter P h , the Lagrange multiplier is corrected, the training process of the support vector machine model is optimized, the performance of the safety distance analysis model is improved, and the trained model is used for real-time safety distance decision.

[0099] Specifically, in the model training process of the optimized support vector machine training and decision unit, a hydraulic pressure influence model is constructed in combination with the hydraulic pressure parameter of the mechanical equipment. The model corrects the Lagrange multiplier with the original Lagrange multiplier, the hydraulic pressure influence coefficient, the current hydraulic pressure parameter and the standard hydraulic pressure reference value as parameters, optimizes the training process of the support vector machine model, so that the model can better adapt to safety distance analysis under different hydraulic pressure conditions, and improve the model performance and decision accuracy.

[0100] As shown in Figure 2 , a digital mechanical equipment safety distance analysis system based on artificial intelligence, the system operation includes the following steps:

[0101] Step one: use the multi-dimensional data acquisition and fusion unit arranged at the key parts of the digital mechanical equipment to acquire multi-source data such as spatial position parameters, running speed parameters, acceleration parameters, and equipment component angle parameters during equipment operation, and fuse the acquired data to form a multi-dimensional original data set;

[0102] Step two: transmit the original data set obtained by the multi-dimensional data acquisition and fusion unit to the data preprocessing and feature extraction unit, standardize the original data, and use the preset feature extraction algorithm to extract the feature vector related to the safety distance analysis of the mechanical equipment from the standardized data;

[0103] Step three: through the binocular vision image acquisition and processing unit equipped with two groups of specific focal length and field angle visual sensors, image acquisition is performed on the surrounding environment of the mechanical equipment, binocular vision image data is acquired, the acquired image data is processed through grayscale, noise reduction, etc., the parallax of each point in the image is calculated through a stereo matching algorithm, and then the three-dimensional space information of the surrounding environment is acquired;

[0104] Step four: the data preprocessing and feature extraction unit extracts the feature vector and the binocular vision image acquisition and processing unit obtains the three-dimensional space information, and transmits them to the optimized support vector machine training and decision unit; based on the optimized support vector machine algorithm, the support vector machine model is trained using historical data, and a safety distance analysis model is constructed; the real-time acquired feature vector and three-dimensional space information are input into the trained safety distance analysis model, and the safety distance decision information of the mechanical equipment and the surrounding objects is output;

[0105] Step five: the safety distance decision information output by the optimized support vector machine training and decision unit is transmitted to the safety distance early warning and control instruction generation unit; according to the preset safety distance threshold, it is judged whether the current running state of the mechanical equipment is in the safety range; when an unsafe state is detected, a corresponding warning signal is generated, and a control instruction is generated according to the safety distance decision information and the control logic of the equipment;

[0106] Step six: the warning signal and control instruction generated by the safety distance early warning and control instruction generation unit are transmitted to the early warning and control execution unit; according to the warning signal, the corresponding warning device is started; according to the control instruction, the running state of the digital mechanical equipment is adjusted.

[0107] The safety distance analysis system of the digital mechanical equipment based on artificial intelligence proposed in the application will be described from three dimensions of data acquisition and processing, intelligent decision and dynamic adaptation.

[0108] In the data acquisition and processing layer, the traditional mechanical equipment safety distance analysis technology often leads to inaccurate analysis due to single data source and simple processing method. The system captures the spatial position, speed, acceleration and other core parameters of the equipment running through the multi-dimensional data acquisition and fusion unit, and uses data fusion technology to build a complete original data set. At the same time, the binocular vision image acquisition and processing unit acquires the surrounding environment image, and through gray processing, noise reduction and stereo matching, accurate three-dimensional space information is obtained. The combination of the two greatly enriches the data dimension. The data preprocessing and feature extraction unit standardizes the multi-source data and extracts the key features, and the optimized support vector machine training and decision unit trains the safety distance analysis model based on this, deeply mines the data correlation, effectively solves the problems of insufficient data processing capacity and low analysis precision of traditional technology, and realizes the comprehensive and accurate perception of the equipment running state and the surrounding environment.

[0109] In terms of intelligent decision-making, the traditional system relies on fixed threshold to judge the safety state, which cannot adapt to complex working condition changes and is prone to misjudgment and missed judgment. The optimized support vector machine training and decision-making unit of the system introduces device parameters such as load weight, temperature, and hydraulic pressure to construct an influence model. During the training process, the model parameters are dynamically adjusted, enabling the system to flexibly adjust the safety distance analysis strategy according to the real-time state of the device. The safety distance warning and control instruction generation unit generates control instructions such as emergency braking and deceleration based on the decision-making information output by the optimized support vector machine and the device braking performance parameters to construct a control instruction generation model. The warning and control execution unit quickly responds to the instructions to achieve intelligent safety management and control, completely changing the status quo of traditional technology lacking intelligent decision-making and adaptive capabilities, and ensuring the safe operation of the device under complex working conditions.

[0110] In the face of complex and variable industrial environments, traditional technology is difficult to effectively cope with the influence of interference factors such as vibration and noise on the accuracy of environmental information collection. The binocular vision image acquisition and processing unit of the system constructs an adaptive noise reduction model based on the device operating frequency and a vibration compensation model based on vibration parameters. The adaptive noise reduction model automatically optimizes the noise reduction strategy based on the device operating frequency to reduce the damage of interference noise to image quality. The vibration compensation model considers factors such as device vibration amplitude and frequency to accurately compensate and correct object distance measurement, ensuring accurate acquisition of three-dimensional spatial information of the surrounding environment. These dynamic environmental adaptation mechanisms significantly enhance the stability and reliability of the system in complex environments, overcoming the drawbacks of traditional technology being restricted by environmental factors and building a strong defense for the safe operation of digital mechanical equipment.

[0111] In the description of the present application, it should be noted that unless otherwise explicitly specified and limited, the terms "arrangement", "installation", "connection", "connection", "fixing" should be broadly understood, for example, it can be fixed connection, or detachable connection, or integral connection; it can be mechanical connection, or electrical connection; it can be directly connected, or indirectly connected through intermediate medium, or the internal communication of two elements. For those skilled in the art, the specific meaning of the above terms in the present application can be understood through specific circumstances.

[0112] Although embodiments of the present application have been shown and described, it will be understood by those skilled in the art that various equivalent changes, modifications, replacements and variations of these embodiments can be made without departing from the principles and spirit of the present application, and the scope of the present application is defined by the appended claims and their equivalent scope.

Claims

1. A digital mechanical equipment safety distance analysis system based on artificial intelligence, characterized in that: include: Multi-dimensional data acquisition and fusion unit: Deployed in key locations of digital mechanical equipment, it collects spatial position, speed, acceleration, and component angle parameters during equipment operation. It uses data fusion technology to integrate multi-source data into an original set and transmits it to the data preprocessing and feature extraction unit. Data preprocessing and feature extraction unit: Receives the original data set, normalizes it to eliminate dimensional differences, uses a preset algorithm to extract feature vectors related to safety distance analysis, and then sends them to the binocular vision image acquisition and processing unit; Binocular vision image acquisition and processing unit: Two sets of visual sensors with specific focal lengths and field of view are configured to collect image data of the surrounding environment. After grayscale conversion and noise reduction, the three-dimensional spatial information is obtained by calculating the parallax using a stereo matching algorithm and transmitted to the optimized support vector machine training and decision-making unit. Optimized support vector machine training and decision-making unit: Receives feature vectors and three-dimensional spatial information, uses historical data to train a model based on an optimized support vector machine algorithm, inputs real-time data into the trained model, outputs safety distance decision information, and sends it to the safety distance warning and control instruction generation unit; Safety distance warning and control instruction generation unit: Receives decision information, determines the equipment operating status based on preset thresholds, generates a warning signal when it is unsafe, and generates control instructions based on the equipment control logic, which are transmitted to the warning and control execution unit; Early warning and control execution unit: receives early warning signals and control instructions, activates the sound and light alarm early warning device, and adjusts the equipment's operating speed, direction, or start and stop according to the instructions.

2. The artificial intelligence-based digital mechanical equipment safety distance analysis system according to claim 1 is characterized in that: The optimized support vector machine training and decision-making unit uses the following optimized support vector machine model to perform safety distance analysis: Where D is the calculated safe distance between the mechanical equipment and surrounding objects; n is the number of training samples; α i is the Lagrange multiplier corresponding to the i-th training sample, and its value range is determined by the optimization algorithm; i is the category label of the i-th training sample, representing whether it is within the safe distance range; K(x i , x) is the kernel function, x i is the feature vector of the i-th training sample, x is the current input feature vector to be analyzed, and the kernel function is used to map data in the low-dimensional space to the high-dimensional space for better classification; b is the offset of the classification hyperplane, which is determined by the training process.

3. The artificial intelligence-based digital mechanical equipment safety distance analysis system according to claim 1 is characterized in that: The binocular vision image acquisition and processing unit adopts the following disparity calculation model in the stereo matching process: Where d is the calculated parallax; f is the focal length of the visual sensor, which is an inherent parameter of the binocular vision sensor; B is the baseline distance between the two visual sensors, which is determined when the system is installed; Z is the actual distance from the object to the visual sensor; at the same time, the calculated parallax is corrected by combining the spatial position parameters P (x, y, z) of the digital mechanical equipment. The corrected parallax d 修正 The calculation formula is: d 修正 =d+Δd(P(x,y,z)) Here, Δd(P(x, y, z)) is the parallax correction amount based on the spatial position parameters P(x, y, z) of the mechanical device.

4. The artificial intelligence-based digital mechanical equipment safety distance analysis system according to claim 1 is characterized in that: The optimized support vector machine training and decision-making unit introduces the load weight W of the mechanical equipment as an influencing factor during the model training process and constructs the following improved optimization objective function: Where ω is the normal vector of the hyperplane; b is the offset of the classification hyperplane; ξ i is the slack variable of the i-th sample; C is the penalty parameter used to balance the complexity of the model and the classification error; λ is the influence coefficient of the load weight W; φ(x i ) is to input sample x i The result of mapping to high-dimensional space through kernel function.

5. The artificial intelligence-based digital mechanical equipment safety distance analysis system according to claim 1 is characterized in that: The binocular vision image acquisition and processing unit uses a noise reduction method based on the operating frequency f of the mechanical equipment when performing noise reduction processing on the acquired image. r Adaptive denoising model: Among them, I 降噪 (x, y) is the pixel value of the denoised image at coordinate (x, y); I(m, n) is the pixel value of the original image at coordinate (m, n); N(x, y) is the neighborhood centered at coordinate (x, y); f(x, y) is the frequency characteristic of the pixel at coordinate (x, y); β is the parameter that controls the weight decay rate.

6. The artificial intelligence-based digital mechanical equipment safety distance analysis system according to claim 1 is characterized in that: The safety distance warning and control instruction generating unit is based on the safety distance decision information D output by the optimized support vector machine and the braking performance parameter B of the mechanical equipment. p , build a control instruction generation model: Among them, C 指令 is the generated control instruction, D 紧急 is the emergency braking safety distance threshold, C 警告 The safety distance warning thresholds are based on the braking performance parameters B of the mechanical equipment. p .

7. The artificial intelligence-based digital mechanical equipment safety distance analysis system according to claim 1 is characterized in that: The optimized support vector machine training and decision-making unit, during model training, combines the influence of the temperature parameter T of the mechanical equipment on the safety distance analysis to construct a temperature correction model: D 修正 =D×(1+γ(T-T0)) Among them, D 修正 is the corrected safety distance value; D is the safety distance value initially calculated by the optimized support vector machine model; γ is the temperature influence coefficient; T is the current temperature parameter of the mechanical equipment; T0 is the standard temperature reference value.

8. The artificial intelligence-based digital mechanical equipment safety distance analysis system according to claim 1 is characterized in that: The binocular vision image acquisition and processing unit, when acquiring three-dimensional spatial information through stereo matching, combines the vibration parameter V of the mechanical equipment to construct a vibration compensation model: With 补偿 =Z+ΔZ(V) Among them, Z 补偿 , is the actual distance from the object to the visual sensor after compensation, is the original distance calculated by parallax, and ΔZ(V) is the distance compensation amount based on the vibration parameter V of the mechanical equipment, which is determined by the pre-established vibration-distance compensation relationship model.

9. The artificial intelligence-based digital mechanical equipment safety distance analysis system according to claim 1, characterized in that: The optimized support vector machine training and decision-making unit combines the hydraulic pressure parameter P of the mechanical equipment during the model training process. h , build the hydraulic pressure influence model: in, is the Lagrange multiplier corresponding to the i-th training sample after correction of the hydraulic pressure effect; α i is the original Lagrange multiplier; μ is the hydraulic pressure influence coefficient; P h is the current hydraulic pressure parameter of the mechanical equipment; P h0 It is the reference value of standard hydraulic pressure.

10. The artificial intelligence-based digital mechanical equipment safety distance analysis system according to any one of claims 1 to 9, characterized in that: The system operation includes the following steps: Step 1: Utilize the multi-dimensional data acquisition and fusion unit installed at the key parts of the digital mechanical equipment to collect multi-source data such as the spatial position parameters, operating speed parameters, acceleration parameters, and angle parameters of each component of the equipment during operation, and fuse the collected data to form a multi-dimensional raw data set; Step 2: The raw data set obtained by the multi-dimensional data acquisition and fusion unit is transmitted to the data preprocessing and feature extraction unit, the raw data is standardized, and a preset feature extraction algorithm is used to extract feature vectors related to the safety distance analysis of mechanical equipment from the standardized data; Step 3: Using a binocular vision image acquisition and processing unit equipped with two sets of visual sensors with specific focal lengths and field of view angles, the system acquires images of the surrounding environment of the mechanical equipment, obtains binocular vision image data, converts the acquired image data into grayscale and performs noise reduction processing, and calculates the parallax of each point in the image using a stereo matching algorithm to obtain three-dimensional spatial information of the surrounding environment. Step 4: The feature vectors extracted by the data preprocessing and feature extraction unit and the three-dimensional spatial information obtained by the binocular vision image acquisition and processing unit are transmitted to the optimized support vector machine training and decision-making unit. Based on the optimized support vector machine algorithm, the support vector machine model is trained using historical data to build a safety distance analysis model. The feature vectors and three-dimensional spatial information obtained in real time are input into the trained safety distance analysis model, and the safety distance decision information between the mechanical equipment and surrounding objects is output; Step 5: The safety distance decision information output by the optimized support vector machine training and decision-making unit is transmitted to the safety distance warning and control instruction generation unit. Based on the preset safety distance threshold, it is determined whether the current operating state of the mechanical equipment is within the safe range. When an unsafe state is detected, a corresponding warning signal is generated, and a control instruction is generated based on the safety distance decision information and the control logic of the equipment. Step 6: The warning signal and control instruction generated by the safety distance warning and control instruction generation unit are transmitted to the warning and control execution unit, the corresponding warning device is activated according to the warning signal, and the operating status of the digital mechanical equipment is adjusted according to the control instruction.

Images