A digital mechanical equipment safety distance analysis system based on artificial intelligence

Abstract

The present invention relates to the field of mechanical equipment safety technology, and discloses a digital mechanical equipment safety distance analysis system based on artificial intelligence, comprising a data acquisition module, a data preprocessing module, a three-dimensional modeling module, an equipment analysis module, a floating data analysis module, an influencing factor analysis module, a safety distance estimation module and a human-computer interaction module; by providing an equipment analysis module, it is beneficial to obtain safety distances in different ways for equipment in normal operation and equipment in a fault state, which is more targeted. For equipment in a normal operating state, the surrounding environment will affect the safety distance of the equipment, while for equipment in a fault state, its own risk factors are the main influencing factors of the equipment safety distance. Therefore, corresponding analysis of equipment in a normal state and equipment in a fault state can improve the accuracy of equipment safety distance analysis.

Description

Technical Field

[0001] The present invention relates to the field of mechanical equipment safety technology, and more specifically to a digital mechanical equipment safety distance analysis system based on artificial intelligence. Background Art

[0002] With the rapid development of industrial automation, mechanical equipment is being used more and more widely in various workplaces; however, during the operation and maintenance of mechanical equipment, if the safety distance between personnel and equipment, and between equipment, is not properly set, it is very easy to cause safety accidents.

[0003] A public document with publication number CN105405248B discloses a large-scale mechanical equipment safety distance alarm system, including an alarm control host, a remote card reader, multiple wireless signal transmission cards and an alarm. The remote card reader is installed on the large-scale mechanical equipment, and the remote card reader and the alarm are connected to the alarm control host. The remote card reader includes a distance measurement module. The remote card reader receives real-time signals from multiple wireless signal transmission cards, measures the distance between the multiple wireless signal transmission cards and the remote card reader through the distance measurement module and transmits it to the alarm control host. When one or more received distance values ​​are less than the preset safety distance, the alarm control host controls the alarm to alarm; an alarm method of the above alarm system is also disclosed, which performs signal marking on potential equipment or personnel with safety hazards, participates in safety distance monitoring, and ensures safety without blind spots.

[0004] However, different equipment has different safety distances. The traditional safety distance monitoring method relies on manual measurement and monitoring, which is not only inefficient, but also has great safety hazards, is time-consuming and labor-intensive, and is difficult to ensure the accuracy and timeliness of the analysis. In view of this, the present invention provides a digital mechanical equipment safety distance analysis system based on artificial intelligence, which uses intelligent algorithms to analyze the safety distance between equipment and the safety distance between equipment and personnel to improve accuracy and timeliness. Summary of the invention

[0005] In order to overcome the above-mentioned defects of the prior art, the present invention provides a digital mechanical equipment safety distance analysis system based on artificial intelligence to solve the problems existing in the above-mentioned background technology.

[0006] The present invention provides the following technical solutions: a digital mechanical equipment safety distance analysis system based on artificial intelligence, comprising a data acquisition module, a data preprocessing module, a three-dimensional modeling module, an equipment analysis module, a floating data analysis module, an influencing factor analysis module, a safety distance estimation module and a human-computer interaction module;

[0007] The data acquisition module is used to collect the operating data and three-dimensional point cloud data of the target mechanical equipment, and transmit them to the data preprocessing module for preprocessing operations;

[0008] The data preprocessing module is used to receive data from the data acquisition module and perform preprocessing operations on the data;

[0009] The three-dimensional modeling module is used to receive the data preprocessed by the three-dimensional point cloud data preprocessing unit, perform three-dimensional modeling, form a three-dimensional point cloud image of the space where the target mechanical equipment is located, and obtain the coordinate point data of the target mechanical equipment;

[0010] The equipment analysis module is used to receive the data preprocessed by the operation data preprocessing unit, analyze the target mechanical equipment based on the probability model, and obtain the operation status of the target mechanical equipment;

[0011] The floating data analysis module is used to receive data from the three-dimensional modeling module and the equipment analysis module, analyze the risk of the target mechanical equipment under the fault state based on the constructed BP neural network model, and obtain the safety distance floating value of the target mechanical equipment;

[0012] The influencing factor analysis module is used to receive data from the equipment analysis module and obtain a safety distance compensation value by analyzing the influencing factors of the target mechanical equipment;

[0013] The safety distance estimation module is used to estimate the optimal safety distance of the target mechanical equipment, obtain the optimal safety distance of the target mechanical equipment in normal operation and fault state, and transmit it to the human-computer interaction module;

[0014] The human-computer interaction module is used to perform human-computer interaction display on the data.

[0015] Preferably, the data acquisition module includes an operation data acquisition unit and a three-dimensional point cloud data acquisition unit. The operation data acquisition unit is used to collect the operation data of the target mechanical equipment and transmit it to the operation data preprocessing unit. The three-dimensional point cloud data acquisition unit is used to collect the three-dimensional point cloud data of the target mechanical equipment and transmit it to the three-dimensional point cloud data preprocessing unit. The operation data of the target mechanical equipment includes the historical operation data and real-time operation data of the target mechanical equipment. The historical operation data includes the operation data of the target mechanical equipment when it was working normally in the past and the operation data of the target mechanical equipment when a failure occurred.

[0016] Preferably, the data preprocessing module includes an operation data preprocessing unit and a three-dimensional point cloud data preprocessing unit. The operation data preprocessing unit is used to preprocess the operation data of the target mechanical equipment, and transmit the obtained operation data to the equipment analysis module; the three-dimensional point cloud data preprocessing unit is used to preprocess the three-dimensional point cloud data of the target mechanical equipment, and transmit the obtained three-dimensional point cloud data to the three-dimensional modeling module.

[0017] Preferably, the specific manner in which the equipment analysis module obtains the operating status of the target mechanical equipment is:

[0018] Input the real-time operation data of the target mechanical equipment into the constructed probability model, obtain the probability of normal operation state and fault state probability of the equipment through the probability model, and select the operation state label corresponding to the value with greater probability as the operation state of the target mechanical equipment;

[0019] If it is in a normal operating state, the data is transmitted to the influencing factor analysis module for further analysis; if it is in a fault state, the data is transmitted to the floating data analysis module for further analysis.

[0020] Preferably, the construction process of the probability model is specifically as follows:

[0021] The historical operation data is marked with a timestamp to form historical operation data of T time periods, and the corresponding historical operation data feature vector is constructed. The operation status label is set and randomly divided into training set data and verification set data; the operation status label is divided into normal operation state A and fault state B;

[0022] Calculate the prior probability of normal operation and fault state of the equipment and the conditional probability of the equipment operation state corresponding to the feature vector of historical operation data in each time period;

[0023] Based on the prior probability and conditional probability, a probability model is constructed. The feature vector X of the historical operation data in the given t-th time period is recorded as a sample, and the probability model calculates ; Among them, P(f|X t ) represents the historical operation data feature vector X in the tth time period t Under the condition of t |f) means that under the condition that the known sample, that is, the historical operation data feature vector of the t-th time period belongs to the operation state label f, the historical operation data feature vector X of the t-th time period is observed. t The probability of the sample belonging to the running state label f ​​is P(X t ) is the evidence probability, indicating that the characteristic vector X of the historical operation data in the tth time period is observed t probability;

[0024] Choose P(f|X t ) The largest running status label is used as the feature vector X of the historical running data in the tth time period tThe predicted output of the running status label is the equipment running status corresponding to the feature vector of the historical running data in the t-th time period; the running status f is divided into normal running status A and fault status B; the probability model is trained using the training set data, and the loss function of the probability model is defined. The cross entropy loss function can be selected, and the constructed probability model is evaluated using the validation set data. The evaluation method is to calculate the value of the loss function. If the value of the loss function does not decrease for λ consecutive times, the probability model at this time is saved as the final probability model; where t=1, 2, 3, …, T.

[0025] Preferably, the specific method for the floating data analysis module to obtain the floating value of the safety distance of the target mechanical equipment is:

[0026] If the equipment analysis module obtains that the operating state of the target mechanical equipment is a fault state, the real-time operating data of the equipment analysis module is received, and the constructed BP neural network model is input to obtain the fault risk level. At the same time, the vibration data in the real-time operating data is input into the trained equipment displacement prediction model to obtain the equipment displacement prediction value; the floating value of the safety distance is obtained through the floating value calculation formula;

[0027] The floating value calculation formula is expressed as:

[0028] ; Among them, Dis_f is the floating value of the safety distance, Dis_b is the standard safety distance, S_yc is the predicted value of the equipment displacement, η is the fault risk index, when the fault risk level is low, the value of η is 1, when the fault risk level is medium, the value of η is 2, when the fault risk level is high, the value of η is 3.

[0029] Preferably, the training method of the equipment displacement prediction model is specifically as follows:

[0030] The vibration data and the corresponding equipment displacement in the operation data of a time period are taken as a group of analysis data, e analysis data are collected in advance, that is, the vibration data and the corresponding equipment displacement in the operation data of e time periods, e is an integer greater than 1, and each group of analysis data is converted into a corresponding set of feature vectors; e=1, 2, 3, ..., T; the device displacement is obtained by: obtaining the coordinate point data of the device at the starting time point of each time period and the coordinate point data of the device at the ending time point, and performing difference calculation on the coordinates of any edge point in the coordinate point data to obtain the device displacement corresponding to the time period;

[0031] Each set of feature vectors is used as the input of the equipment displacement prediction model. The equipment displacement prediction model uses a predicted equipment displacement value corresponding to each set of analysis data as the output, and uses the actual equipment displacement corresponding to each set of analysis data as the prediction target. The actual equipment displacement is obtained from the equipment displacement corresponding to each set of analysis data collected in advance. Minimizing the sum of prediction errors in all analysis data is used as the training target. The formula for the prediction error is expressed as: , where ε p is the prediction error, p is the group number of the eigenvector corresponding to the analyzed data, γ p is the predicted value of equipment displacement corresponding to the pth group of analysis data, μ p The actual equipment displacement corresponding to the pth group of analysis data is used to train the equipment displacement prediction model until the sum of the prediction errors reaches convergence and the training is stopped.

[0032] Preferably, the specific method for the influencing factor analysis module to obtain the safety distance compensation value is:

[0033] Input the environmental data in the real-time operation data into the weight distribution model to obtain the weight value of each environmental factor in the environmental data, and use the compensation formula to obtain the safety distance compensation value;

[0034] The compensation formula is expressed as: , where Dis_sb is the safety distance compensation value, k j is the weight value of the jth environmental factor, J is the number of environmental factors in the real-time operation data, and Dis_b is the standard safety distance.

[0035] Preferably, the construction process of the weight distribution model is specifically as follows:

[0036] Each environmental factor and the corresponding safety distance in the environmental data when the target mechanical equipment is working normally in the historical operation data are taken as a group of research data, and g groups of research data are collected in advance, that is, the operation data, environmental data and corresponding safety distance of the equipment when it is working normally in g time periods;

[0037] The corresponding weight distribution values ​​are set for the g groups of research data, where g is an integer greater than 1, and the research data and the corresponding weight distribution values ​​are converted into a corresponding set of feature vectors; the weight distribution values ​​reflect the degree of influence of different environmental factors in the environmental data on the setting of the equipment safety distance; the corresponding weight distribution values ​​are set for the g groups of research data in turn;

[0038] Each group of feature vectors is used as the input of the weight allocation model. The weight allocation model takes a group of predicted weight values ​​corresponding to each group of analysis data as output, and takes the actual weight allocation values ​​corresponding to each group of analysis data as prediction targets. The actual weight allocation values ​​are the pre-collected weight allocation values ​​corresponding to the research data. The training goal is to minimize the sum of prediction errors of all research data. The formula for the prediction error is the same as that of the equipment displacement prediction model, and training is stopped when the sum of the prediction errors converges.

[0039] Preferably, the specific manner in which the safety distance estimation module obtains the optimal safety distance of the target mechanical equipment in a normal operating state and a fault state is:

[0040] When the equipment is in normal operation, the safety distance floating value is used as the optimal safety distance;

[0041] When the device is in a fault state, the optimal safety distance is: ; Among them, Dis_zj is the optimal safety distance, Dis_sb is the safety distance compensation value, and Dis_b is the standard safety distance.

[0042] Technical effects and advantages of the present invention:

[0043] (1) The present invention is provided with a device analysis module, which is conducive to obtaining the safety distance of the device in normal operation and the device in fault state in different ways, which is more targeted. For the device in normal operation, the surrounding environment will affect the safety distance of the device, while for the device in fault state, its own risk factors are the main influencing factors of the device safety distance. Therefore, corresponding analysis of the device in normal state and the device in fault state can improve the accuracy of the analysis of the device safety distance.

[0044] (2) The present invention is provided with a floating data analysis module, which is conducive to analyzing the floating situation of the safety distance of the equipment in the fault state. Since the risk of the equipment is increased when it is in the fault state and the equipment may be displaced, the safety distance of the equipment will fluctuate due to different risks. The higher the risk in the equipment fault state, the more it is necessary to increase the safety distance to ensure the safety of the equipment and personnel. By calculating the floating value of the safety distance in the equipment fault state to obtain a more accurate safety distance, the safety of the equipment and personnel is improved.

[0045] (3) The present invention is provided with an influencing factor analysis module, which is beneficial to effectively capture complex data relationships by utilizing the nonlinear and multimodal data modeling capabilities of the neural network model; based on the parallel computing processing mechanism, the computing efficiency is improved, and the weight value of each influencing factor is obtained, thereby obtaining the safety distance compensation value, that is, the safety distance value that needs to be compensated under the influence of the influencing factors, thereby improving the accuracy of the safety distance analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 This is a flow chart of the artificial intelligence-based digital mechanical equipment safety distance analysis system of the present invention.

[0047] Figure 2 It is a structural diagram of the data acquisition module of the present invention.

[0048] Figure 3 This is a structural diagram of the data preprocessing module of the present invention. DETAILED DESCRIPTION

[0049] The technical solution of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the present invention. In addition, the forms of the various structures recorded in the following embodiments are merely illustrative. The digital mechanical equipment safety distance analysis system based on artificial intelligence involved in the present invention is not limited to the various structures recorded in the following embodiments. All other implementations obtained by ordinary technicians in this field without making creative work belong to the scope of protection of the present invention.

[0050] like Figure 1 As shown, the present invention provides a digital mechanical equipment safety distance analysis system based on artificial intelligence, including a data acquisition module, a data preprocessing module, a three-dimensional modeling module, an equipment analysis module, a floating data analysis module, an influencing factor analysis module, a safety distance estimation module and a human-computer interaction module;

[0051] The data acquisition module is used to collect the operation data and three-dimensional point cloud data of the target mechanical equipment, and transmit them to the data preprocessing module for preprocessing operations; the data acquisition module includes an operation data acquisition unit and a three-dimensional point cloud data acquisition unit, the operation data acquisition unit is used to collect the operation data of the target mechanical equipment and transmit it to the operation data preprocessing unit, and the three-dimensional point cloud data acquisition unit is used to collect the three-dimensional point cloud data of the target mechanical equipment and transmit it to the three-dimensional point cloud data preprocessing unit; the target mechanical equipment is a mechanical equipment that needs to be analyzed for a safe distance, and the operation data of the target mechanical equipment includes the historical operation data and real-time operation data of the target mechanical equipment, the historical operation data includes the operation data of the target mechanical equipment when it is working normally in the past and the operation data of the target mechanical equipment when it fails, the operation data includes but is not limited to the voltage, current, power, vibration, energy consumption and environment data when the equipment is running, and the historical operation data of the target mechanical equipment can be obtained by obtaining the operation data of multiple identical mechanical equipment when they are working normally and the operation data when they fail by using network crawler technology; the three-dimensional point cloud data includes the three-dimensional point cloud data of the target mechanical equipment itself and the three-dimensional point cloud data of surrounding objects;

[0052] The data preprocessing module is used to receive data from the data acquisition module and perform preprocessing operations on the data; the data preprocessing module includes an operation data preprocessing unit and a 3D point cloud data preprocessing unit, the operation data preprocessing unit is used to preprocess the operation data of the target mechanical equipment, and transmit the obtained operation data to the equipment analysis module; the 3D point cloud data preprocessing unit is used to preprocess the 3D point cloud data of the target mechanical equipment, and transmit the obtained 3D point cloud data to the 3D modeling module; the preprocessing operations include but are not limited to cleaning, denoising and normalization operations on the data;

[0053] The 3D modeling module is used to receive the data preprocessed by the 3D point cloud data preprocessing unit, perform 3D modeling, form a 3D point cloud image of the space where the target mechanical equipment is located, and obtain the coordinate point data of the target mechanical equipment; the coordinate point data is the coordinate point of the edge point of the contact surface between the equipment and the ground; its purpose is to accurately obtain the target mechanical equipment and the surrounding environment and coordinate data, thereby laying a foundation for the subsequent analysis of the floating value of the safety distance of the equipment, and at the same time, it can re-estimate the safety distance of the equipment that has been replaced, and can prepare effective data for the safety distance of different equipment at different positions; some equipment will produce slight displacement due to vibration when a fault occurs, and the safety distance will change at this time, that is, the coordinate points of the safety range will change. If this continues for a long time, if it cannot be discovered in time, it will pose a threat to the safety of equipment and personnel;

[0054] The equipment analysis module is used to receive the data preprocessed by the operation data preprocessing unit, analyze the target mechanical equipment based on the probability model, and obtain the operating status of the target mechanical equipment; the operating status of the target mechanical equipment includes two types: normal operating status and fault status; its purpose is to use different methods to obtain the safety distance for the equipment in normal operating state and the equipment in fault state, which is more targeted. For the equipment in normal operating state, the surrounding environment will affect the safety distance of the equipment, and for the equipment in fault state, its own risk factor is the main influencing factor of the equipment safety distance. Therefore, corresponding analysis of the equipment in normal state and the equipment in fault state can improve the accuracy of the equipment safety distance analysis;

[0055] The floating data analysis module is used to receive data from the three-dimensional modeling module and the equipment analysis module, analyze the risks of the target mechanical equipment under the fault state based on the constructed BP neural network model, and obtain the floating value of the safety distance of the target mechanical equipment; its purpose is to analyze the floating situation of the safety distance of the equipment in the fault state. Since the risk of the equipment increases when it is in the fault state and the equipment may be displaced, the safety distance of the equipment will fluctuate due to different risks. The higher the risk in the equipment fault state, the more it is necessary to increase the safety distance to ensure the safety of the equipment and personnel. By calculating the floating value of the safety distance in the equipment fault state to obtain a more accurate safety distance, the safety of the equipment and personnel is improved.

[0056] The influencing factor analysis module is used to receive data from the equipment analysis module, and obtain the safety distance compensation value by analyzing the influencing factors of the target mechanical equipment; its purpose is to effectively capture complex data relationships by utilizing the nonlinear and multimodal data modeling capabilities of the neural network model; based on the parallel computing processing mechanism, the computing efficiency is improved, and the weight value of each influencing factor is obtained, thereby obtaining the safety distance compensation value, that is, the safety distance value that needs to be compensated under the influence of the influencing factors, and improving the accuracy of the safety distance analysis;

[0057] The safety distance estimation module is used to estimate the optimal safety distance of the target mechanical equipment, obtain the optimal safety distance of the target mechanical equipment under normal operation and fault state, and transmit it to the human-computer interaction module; its purpose is to estimate the optimal safety distance under normal operation and fault state respectively, which is more targeted, and the estimation result of the safety distance is more accurate, which can effectively ensure the safety of equipment and personnel;

[0058] The human-computer interaction module is used to perform human-computer interaction display on the data.

[0059] In this embodiment, it should be specifically explained that the specific manner in which the equipment analysis module obtains the operating status of the target mechanical equipment is:

[0060] Input the real-time operation data of the target mechanical equipment into the constructed probability model, obtain the probability of normal operation state and fault state probability of the equipment through the probability model, and select the operation state label corresponding to the value with greater probability as the operation state of the target mechanical equipment;

[0061] If it is a normal operating state, the data is transmitted to the influencing factor analysis module for further analysis; if it is a fault state, the data is transmitted to the floating data analysis module for further analysis;

[0062] The construction process of the probability model is specifically as follows:

[0063] The historical operation data is marked with a timestamp to form historical operation data of T time periods, and a corresponding historical operation data feature vector is constructed, and an operation status label is set, and the data is randomly divided into training set data and verification set data. The ratio of the training set data to the verification set data can be set at a reasonable ratio, but the ratio of the training set data is greater than the ratio of the verification set data. In this embodiment, the ratio of the training set data to the verification set data is seven to three; the operation status label is divided into normal operating state A and fault state B; the equipment operation status corresponding to the historical operation data of each time period can be displayed. If the historical operation data of the time period is the operation data when the equipment is working normally, the operation status label is the normal operating state. If the historical operation data of the time period is the operation data when the equipment fails, the operation status label is the fault state;

[0064] Calculate the prior probability of normal operation and fault state of the equipment and the conditional probability of the equipment operation state corresponding to the feature vector of historical operation data in each time period; the prior probability is estimated by the proportion of samples corresponding to the operation state in the training set, and the conditional probability follows Gaussian distribution;

[0065] Based on the prior probability and conditional probability, a probability model is constructed. The feature vector X of the historical operation data in the given t-th time period is recorded as a sample, and the probability model calculates ; Among them, P(f|X t ) represents the historical operation data feature vector X in the tth time period t Under the condition of t |f) means that under the condition that the known sample, that is, the historical operation data feature vector of the t-th time period belongs to the operation state label f, the historical operation data feature vector X of the t-th time period is observed. t The probability of the sample belonging to the running state label f ​​is the conditional probability distribution estimated based on the training set data. P(X) is the prior probability, which indicates the probability that the sample belongs to the running state label f. It does not consider any feature information and is calculated based on the ratio of samples in the normal running state to the fault state. t ) is the evidence probability, indicating that the characteristic vector X of the historical operation data in the tth time period is observed t The probability of is independent of the running state. Since it is the same constant value for all running states, it can be ignored when comparing the probabilities of two running states.

[0066] Choose P(f|X t ) The largest running status label is used as the feature vector X of the historical running data in the tth time period tThe predicted output of the running status label is the equipment running status corresponding to the feature vector of the historical running data in the t-th time period; the running status f is divided into normal running status A and fault status B; the probability model is trained using the training set data, and the loss function of the probability model is defined. The cross entropy loss function can be selected, and the constructed probability model is evaluated using the validation set data. The evaluation method is to calculate the value of the loss function. If the value of the loss function does not decrease for λ consecutive times, the probability model at this time is saved as the final probability model; where t=1, 2, 3, …, T.

[0067] In this embodiment, it should be specifically explained that the specific method for the floating data analysis module to obtain the floating value of the safety distance of the target mechanical equipment is:

[0068] If the equipment analysis module obtains that the operating state of the target mechanical equipment is a fault state, the real-time operating data of the equipment analysis module is received, and the constructed BP neural network model is input to obtain the fault risk level. At the same time, the vibration data in the real-time operating data is input into the trained equipment displacement prediction model to obtain the equipment displacement prediction value; the floating value of the safety distance is obtained through the floating value calculation formula;

[0069] The floating value calculation formula is expressed as:

[0070] ; Where Dis_f is the floating value of the safety distance, Dis_b is the standard safety distance, S_yc is the predicted value of the equipment displacement, η is the fault risk index, when the fault risk level is low, the value of η is 1, when the fault risk level is medium, the value of η is 2, when the fault risk level is high, the value of η is 3;

[0071] The standard safety distance is the safety distance of the target mechanical equipment specified in the instructions;

[0072] The BP neural network model includes an input layer, a hidden layer, an activation function and an output layer. The input layer is used to input real-time operation data. The output layer has three nodes, which correspond to three fault risk levels, namely low, medium and high. The activation function selects either a Sigmoid function or a ReLU function to introduce nonlinear factors and enhance the expression ability of the network. The weights and biases of the BP neural network are randomly initialized and adjusted during the training process.

[0073] The operation data of the target mechanical equipment when the failure occurs in the historical operation data and the corresponding failure risk level are used as model data, which are divided into model training data and model verification data. The ratio of the model training data to the model verification data can be set by technicians in this field within a reasonable range. In this embodiment, 70% is still selected as model training data and 30% is selected as model verification data. The failure risk level corresponding to the operation data of the mechanical equipment when the failure occurs in the historical operation data can be judged by technicians in this field according to the failure situation at that time, and an expert group can be used to judge the failure risk level.

[0074] Initialize the BP neural network model, input the model training data into the BP neural network model, and the input data is forward propagated through each layer of the BP neural network. The neurons in each layer receive weighted inputs from the neurons in the previous layer and are processed through the activation function to obtain the output of this layer. This process continues until the output layer to obtain the predicted output. According to the difference between the network predicted output and the actual target output, the error is calculated using the error function, which is the mean square error. The error is back-propagated from the output layer to the hidden layer layer by layer until the input layer. The neurons in each layer receive error signals from the neurons in the next layer. The error signal indicates the contribution of the neurons in this layer to the final error. The chain method is used to calculate the error. The gradient of the error with respect to the weights of each layer is calculated. The gradient indicates the degree of influence of the weight change on the error reduction. The gradient descent algorithm is used to update the weights and biases of the network based on the calculated gradient information. The update direction of the weights and biases is the opposite direction of the gradient, and its purpose is to reduce the error. The new weight is the old weight minus the product of the learning rate and the gradient. The learning rate is a hyperparameter used to control the step size of the weight update. Forward propagation and backward propagation are repeated until the stopping condition is met. The stopping condition can be either reaching the preset maximum number of iterations or the error is less than the set threshold. Regularization can be used in the training process to prevent overfitting. The verification data is then used to verify and evaluate the BP neural network model.

[0075] In this embodiment, it should be specifically explained that the training method of the equipment displacement prediction model is specifically:

[0076] The vibration data and the corresponding equipment displacement in the operation data of a time period are taken as a group of analysis data, e analysis data are collected in advance, that is, the vibration data and the corresponding equipment displacement in the operation data of e time periods, e is an integer greater than 1, and each group of analysis data is converted into a corresponding set of feature vectors; e=1, 2, 3, ..., T; the device displacement is obtained by: obtaining the coordinate point data of the device at the starting time point of each time period and the coordinate point data of the device at the ending time point, and performing difference calculation on the coordinates of any edge point in the coordinate point data to obtain the device displacement corresponding to the time period;

[0077] Each set of feature vectors is used as the input of the equipment displacement prediction model. The equipment displacement prediction model uses a predicted equipment displacement value corresponding to each set of analysis data as the output, and uses the actual equipment displacement corresponding to each set of analysis data as the prediction target. The actual equipment displacement is obtained from the equipment displacement corresponding to each set of analysis data collected in advance. Minimizing the sum of prediction errors in all analysis data is used as the training target. The formula for the prediction error is expressed as: , where ε p is the prediction error, p is the group number of the eigenvector corresponding to the analyzed data, γ p is the predicted value of equipment displacement corresponding to the pth group of analysis data, μ p The actual equipment displacement corresponding to the p-th group of analysis data is used to train the equipment displacement prediction model until the sum of the prediction errors reaches convergence and the training is stopped;

[0078] The equipment displacement prediction model is specifically a deep network model.

[0079] In this embodiment, it should be specifically explained that the specific method for the influencing factor analysis module to obtain the safety distance compensation value is:

[0080] Input the environmental data in the real-time operation data into the weight distribution model to obtain the weight value of each environmental factor in the environmental data, and use the compensation formula to obtain the safety distance compensation value;

[0081] The compensation formula is expressed as: , where Dis_sb is the safety distance compensation value, k j is the weight value of the jth environmental factor, and J is the number of environmental factors in the real-time operation data; for example, if the environmental data in the real-time operation data only have two data items, temperature and humidity, the number of environmental factors is 2, that is, J=2; the environmental data in the real-time operation data can be set by the technicians in this field as the detection data items of the environmental factors according to the equipment conditions. If the equipment is located in an environment with high dust concentration and humidity, or the equipment is sensitive to humidity and dust, the environmental data can be set as dust concentration and humidity;

[0082] The construction process of the weight distribution model is specifically as follows:

[0083] Each environmental factor and the corresponding safety distance in the environmental data when the target mechanical equipment in the historical operation data is working normally are taken as a group of research data, and g groups of research data are collected in advance, that is, the operation data, environmental data and corresponding safety distance of the equipment when it is working normally in g time periods; since environmental factors will affect the safety distance of the equipment, for example, in a humid, high-dust or corrosive gas environment, the safety distance needs to be increased to ensure the normal operation of the equipment and the safety of personnel, so it is necessary to study the impact of environmental factors to improve the accuracy of the safety distance and maximize the safety of equipment and personnel; the environmental factors are all environmental factors, including but not limited to temperature, humidity, dust content and corrosive gas content; if all environmental factors are N, then J∈[1,N];

[0084] Corresponding weight distribution values ​​are set for g groups of research data, where g is an integer greater than 1, and the research data and the corresponding weight distribution values ​​are converted into a corresponding set of feature vectors; the weight distribution values ​​corresponding to the research data are obtained by technical personnel in this field during the environmental data impact assessment process corresponding to the historical equipment safety distance setting, and g groups of research data are collected. Under the conditions of each group of research data, the changes in real-time equipment environmental data are comprehensively studied, and the degree of influence on the equipment safety distance setting is combined with actual experience to set the corresponding weight distribution values; the weight distribution values ​​reflect the degree of influence of different environmental factors in the environmental data on the equipment safety distance setting; the corresponding weight distribution values ​​are set for the g groups of research data in turn;

[0085] Each group of feature vectors is used as the input of the weight allocation model. The weight allocation model uses a group of predicted weight values ​​corresponding to each group of analysis data as output, and uses the actual weight allocation values ​​corresponding to each group of analysis data as the prediction target. The actual weight allocation values ​​are the pre-collected weight allocation values ​​corresponding to the research data. Minimizing the sum of the prediction errors of all research data is used as the training target. The formula of the prediction error is the same as that of the equipment displacement prediction model, and the training is stopped until the sum of the prediction errors reaches convergence.

[0086] The weight distribution model includes an input layer, a hidden layer and an output layer; each hidden layer includes multiple neurons, each neuron is connected to the neurons in the next layer, and the connection contains weights, which determine the importance and influence of data transmission in the neural network; an activation function is applied to each neuron between the hidden layer and the output layer, and the activation function is mapped into nonlinearity, allowing the network to learn more complex patterns and features.

[0087] In this embodiment, it should be specifically explained that the specific manner in which the safety distance estimation module obtains the optimal safety distance of the target mechanical equipment in the normal operating state and the fault state is:

[0088] When the equipment is in normal operation, the safety distance floating value is used as the optimal safety distance;

[0089] When the device is in a fault state, the optimal safety distance is: ; Among them, Dis_zj is the optimal safety distance.

[0090] In this embodiment, it should be specifically explained that the difference between this embodiment and the prior art lies in that this embodiment has an equipment analysis module, and adopts different methods to obtain the safety distance for the equipment in normal operation and the equipment in fault state, which is more targeted. For the equipment in normal operation, the surrounding environment will affect the safety distance of the equipment, while for the equipment in fault state, its own risk factor is the main influencing factor of the equipment safety distance. Therefore, the corresponding analysis of the equipment in normal state and the equipment in fault state can improve the accuracy of the equipment safety distance analysis; it has a floating data analysis module to analyze the floating situation of the safety distance of the equipment in fault state. Since the risk of the equipment increases when it is in fault state, And equipment displacement may occur, so the safety distance of the equipment will fluctuate due to different risks. The higher the risk in the equipment failure state, the more the safety distance needs to be increased to ensure the safety of equipment and personnel. By calculating the floating value of the safety distance in the equipment failure state to obtain a more accurate safety distance, the safety of equipment and personnel is improved; it has an influencing factor analysis module, which effectively captures complex data relationships by utilizing the nonlinear and multimodal data modeling capabilities of the neural network model; based on the parallel computing processing mechanism, it improves computing efficiency, obtains the weight value of each influencing factor, and thus obtains the safety distance compensation value, that is, the safety distance value that needs to be compensated under the influence of the influencing factors, thereby improving the accuracy of safety distance analysis.

[0091] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

[0092] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A digital mechanical equipment safety distance analysis system based on artificial intelligence, characterized by: It includes data acquisition module, data preprocessing module, 3D modeling module, equipment analysis module, floating data analysis module, influencing factor analysis module, safety distance estimation module and human-computer interaction module; The data acquisition module is used to collect the operating data and three-dimensional point cloud data of the target mechanical equipment, and transmit them to the data preprocessing module for preprocessing operations; The data preprocessing module is used to receive data from the data acquisition module and perform preprocessing operations on the data; The three-dimensional modeling module is used to receive the data preprocessed by the three-dimensional point cloud data preprocessing unit, perform three-dimensional modeling, form a three-dimensional point cloud image of the space where the target mechanical equipment is located, and obtain the coordinate point data of the target mechanical equipment; The equipment analysis module is used to receive the data preprocessed by the operation data preprocessing unit, analyze the target mechanical equipment based on the probability model, and obtain the operation status of the target mechanical equipment; The floating data analysis module is used to receive data from the three-dimensional modeling module and the equipment analysis module, analyze the risk of the target mechanical equipment under the fault state based on the constructed BP neural network model, and obtain the safety distance floating value of the target mechanical equipment; The influencing factor analysis module is used to receive data from the equipment analysis module and obtain a safety distance compensation value by analyzing the influencing factors of the target mechanical equipment; The safety distance estimation module is used to estimate the optimal safety distance of the target mechanical equipment, obtain the optimal safety distance of the target mechanical equipment in normal operation and fault state, and transmit it to the human-computer interaction module; The human-computer interaction module is used to perform human-computer interaction display on the data; The specific method for the floating data analysis module to obtain the floating value of the safety distance of the target mechanical equipment is: If the equipment analysis module obtains that the operating state of the target mechanical equipment is a fault state, the real-time operating data of the equipment analysis module is received, and the constructed BP neural network model is input to obtain the fault risk level. At the same time, the vibration data in the real-time operating data is input into the trained equipment displacement prediction model to obtain the equipment displacement prediction value; the floating value of the safety distance is obtained through the floating value calculation formula; The floating value calculation formula is expressed as: Dis_f=(Dis_b+S_yc) η ; Where Dis_f is the floating value of the safety distance, Dis_b is the standard safety distance, S_yc is the predicted value of the equipment displacement, η is the fault risk index, when the fault risk level is low, the value of η is 1, when the fault risk level is medium, the value of η is 2, when the fault risk level is high, the value of η is 3; The specific method for the influencing factor analysis module to obtain the safety distance compensation value is: Input the environmental data in the real-time operation data into the weight distribution model to obtain the weight value of each environmental factor in the environmental data, and use the compensation formula to obtain the safety distance compensation value; The compensation formula is expressed as: Among them, Dis_sb is the safety distance compensation value, k j is the weight value of the jth environmental factor, J is the number of environmental factors in the real-time operation data, and Dis_b is the standard safety distance; The specific method for the safety distance estimation module to obtain the optimal safety distance of the target mechanical equipment in normal operation and fault state is: When the equipment is in normal operation, the safety distance floating value is used as the optimal safety distance; When the device is in a fault state, the optimal safety distance is: Dis_zj = Dis_sb + Dis_b; where Dis_zj is the optimal safety distance, Dis_sb is the safety distance compensation value, and Dis_b is the standard safety distance.

2. According to the artificial intelligence-based digital mechanical equipment safety distance analysis system of claim 1, it is characterized by: The data acquisition module includes an operation data acquisition unit and a three-dimensional point cloud data acquisition unit. The operation data acquisition unit is used to collect the operation data of the target mechanical equipment and transmit it to the operation data preprocessing unit. The three-dimensional point cloud data acquisition unit is used to collect the three-dimensional point cloud data of the target mechanical equipment and transmit it to the three-dimensional point cloud data preprocessing unit. The operation data of the target mechanical equipment includes the historical operation data and real-time operation data of the target mechanical equipment. The historical operation data includes the operation data of the target mechanical equipment when it was working normally in the past and the operation data of the target mechanical equipment when a failure occurred.

3. According to claim 2, a digital mechanical equipment safety distance analysis system based on artificial intelligence is characterized in that: The data preprocessing module includes an operation data preprocessing unit and a three-dimensional point cloud data preprocessing unit. The operation data preprocessing unit is used to preprocess the operation data of the target mechanical equipment, and transmit the obtained operation data that can be directly used to the equipment analysis module; the three-dimensional point cloud data preprocessing unit is used to preprocess the three-dimensional point cloud data of the target mechanical equipment, and transmit the obtained three-dimensional point cloud data that can be directly used to the three-dimensional modeling module.

4. According to claim 3, a digital mechanical equipment safety distance analysis system based on artificial intelligence is characterized in that: The specific method for the equipment analysis module to obtain the operating status of the target mechanical equipment is: Input the real-time operation data of the target mechanical equipment into the constructed probability model, obtain the probability of normal operation state and fault state probability of the equipment through the probability model, and select the operation state label corresponding to the value with greater probability as the operation state of the target mechanical equipment; If it is in a normal operating state, the data is transmitted to the influencing factor analysis module for further analysis; if it is in a fault state, the data is transmitted to the floating data analysis module for further analysis.

5. The artificial intelligence-based digital mechanical equipment safety distance analysis system according to claim 4 is characterized in that: The construction process of the probability model is specifically as follows: The historical operation data is marked with a timestamp to form historical operation data of T time periods, and the corresponding historical operation data feature vector is constructed. The operation status label is set and randomly divided into training set data and verification set data; the operation status label is divided into normal operation state A and fault state B; Calculate the prior probability of normal operation and fault state of the equipment and the conditional probability of the equipment operation state corresponding to the feature vector of historical operation data in each time period; Based on the prior probability and conditional probability, a probability model is constructed. The feature vector X of the historical operation data in the given t-th time period is recorded as a sample, and the probability model calculates Among them, P(f|X t ) represents the historical operation data feature vector X in the tth time period t Under the condition of t |f) means that under the condition that the known sample, that is, the historical operation data feature vector of the t-th time period belongs to the operation state label f, the historical operation data feature vector X of the t-th time period is observed. t The probability of the sample belonging to the running state label f ​​is P(X t ) is the evidence probability, indicating that the characteristic vector X of the historical operation data in the tth time period is observed t The probability of Choose P(f|X t ) The largest running status label is used as the feature vector X of the historical running data in the tth time period t The predicted output of the running status label is the equipment running status corresponding to the feature vector of the historical running data in the t-th time period; the running status f is divided into normal running status A and fault status B; the probability model is trained using the training set data, and the loss function of the probability model is defined. The cross entropy loss function can be selected, and the constructed probability model is evaluated using the validation set data. The evaluation method is to calculate the value of the loss function. If the value of the loss function does not decrease for λ consecutive times, the probability model at this time is saved as the final probability model; where t = 1, 2, 3, …, T.

6. The artificial intelligence-based digital mechanical equipment safety distance analysis system according to claim 1, characterized in that: The training method of the equipment displacement prediction model is specifically as follows: The vibration data and the corresponding equipment displacement in the operation data of a time period are taken as a group of analysis data, e analysis data are collected in advance, that is, the vibration data and the corresponding equipment displacement in the operation data of e time periods, e is an integer greater than 1, and each group of analysis data is converted into a corresponding set of feature vectors; e=1, 2, 3, ..., T; the equipment displacement is obtained by: obtaining the coordinate point data of the equipment at the starting time point of each time period and the coordinate point data of the equipment at the ending time point, and performing difference calculation on the coordinates of any edge point in the coordinate point data to obtain the equipment displacement corresponding to the time period; Each set of feature vectors is used as the input of the equipment displacement prediction model. The equipment displacement prediction model takes a predicted equipment displacement value corresponding to each set of analysis data as the output, and takes the actual equipment displacement corresponding to each set of analysis data as the prediction target. The actual equipment displacement is obtained from the equipment displacement corresponding to each set of analysis data collected in advance. The training goal is to minimize the sum of the prediction errors in all analysis data. The formula for the prediction error is expressed as: ε p =γ p -μ p , where ε p is the prediction error, p is the group number of the eigenvector corresponding to the analyzed data, γ p is the predicted value of equipment displacement corresponding to the pth group of analysis data, μ p The actual equipment displacement corresponding to the pth group of analysis data is used to train the equipment displacement prediction model until the sum of the prediction errors reaches convergence and the training is stopped.

7. The digital mechanical equipment safety distance analysis system based on artificial intelligence according to claim 1 is characterized by: The construction process of the weight distribution model is specifically as follows: Each environmental factor and the corresponding safety distance in the environmental data when the target mechanical equipment is working normally in the historical operation data are taken as a group of research data, and g groups of research data are collected in advance, that is, the operation data, environmental data and corresponding safety distance of the equipment when it is working normally in g time periods; The corresponding weight distribution values ​​are set for the g groups of research data, where g is an integer greater than 1, and the research data and the corresponding weight distribution values ​​are converted into a corresponding set of feature vectors; the weight distribution values ​​reflect the degree of influence of different environmental factors in the environmental data on the setting of the equipment safety distance; the corresponding weight distribution values ​​are set for the g groups of research data in turn; Each group of feature vectors is used as the input of the weight allocation model. The weight allocation model takes a group of predicted weight values ​​corresponding to each group of analysis data as output, and takes the actual weight allocation values ​​corresponding to each group of analysis data as prediction targets. The actual weight allocation values ​​are the pre-collected weight allocation values ​​corresponding to the research data. The training goal is to minimize the sum of prediction errors of all research data. The formula for the prediction error is the same as that of the equipment displacement prediction model, and training is stopped when the sum of the prediction errors converges.

Images